{"id":31457,"date":"2021-12-30T23:09:18","date_gmt":"2021-12-30T15:09:18","guid":{"rendered":"https:\/\/swarma.org\/?p=31457"},"modified":"2021-12-30T23:09:18","modified_gmt":"2021-12-30T15:09:18","slug":"%e9%95%bf%e6%96%87%e7%bb%bc%e8%bf%b0%ef%bc%9a%e7%bb%99%e7%94%9f%e7%89%a9%e5%ad%a6%e5%ae%b6%e7%9a%84%e6%9c%ba%e5%99%a8%e5%ad%a6%e4%b9%a0%e6%8c%87%e5%8d%97","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=31457","title":{"rendered":"\u957f\u6587\u7efc\u8ff0\uff1a\u7ed9\u751f\u7269\u5b66\u5bb6\u7684\u673a\u5668\u5b66\u4e60\u6307\u5357"},"content":{"rendered":"
\n
\n
\n
\n
\n

<\/p>\n<\/section>\n

\n
\n
\n
\n
\n
\n

<\/section>\n<\/section>\n
\n
\n

\u5bfc\u8bed<\/strong><\/p>\n<\/section>\n<\/section>\n

\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\n
\n
\n

<\/strong>\u8fc7\u53bb\u51e0\u5341\u5e74\uff0c\u751f\u7269\u6570\u636e\u96c6\u7684\u89c4\u6a21\u4e0e\u590d\u6742\u6027\u5927\u5e45\u589e\u957f\uff0c\u8fd9\u4f7f\u5f97\u673a\u5668\u5b66\u4e60\u8d8a\u6765\u8d8a\u591a\u5730\u7528\u4e8e\u4e3a\u6f5c\u5728\u751f\u7269\u8fc7\u7a0b\u6784\u5efa\u4fe1\u606f\u4e0e\u9884\u6d4b\u6a21\u578b\u3002\u6240\u6709\u673a\u5668\u5b66\u4e60\u6280\u672f\u90fd\u5728\u8ba9\u6a21\u578b\u4e0e\u6570\u636e\u76f8\u5339\u914d\uff1b\u7136\u800c\uff0c\u5177\u4f53\u7684\u65b9\u6cd5\u591a\u79cd\u591a\u6837\uff0c\u4e4d\u4e00\u770b\u4f3c\u4e4e\u4ee4\u4eba\u773c\u82b1\u7f2d\u4e71\u3002\u5bf9\u4e8e\u4e0d\u540c\u7c7b\u578b\u7684\u751f\u7269\u6570\u636e\uff0c\u8be5\u5982\u4f55\u9009\u62e9\u7279\u5b9a\u7684\u673a\u5668\u5b66\u4e60\u6280\u672f\uff1f<\/span><\/strong><\/span><\/p>\n


<\/strong><\/p>\n

2021\u5e749\u6708\uff0c\u53d1\u8868\u5728 Nature Reviews Molecular Cell Biology<\/em> \u4e0a\u7684\u7efc\u8ff0\u6587\u7ae0\u201c\u7ed9\u751f\u7269\u5b66\u5bb6\u7684\u673a\u5668\u5b66\u4e60\u6307\u5357\u201d\uff0c\u5411\u8bfb\u8005\u7b80\u8981\u4ecb\u7ecd\u4e86\u4e00\u4e9b\u5173\u952e\u7684\u673a\u5668\u5b66\u4e60\u6280\u672f\uff1a\u65e2\u5305\u62ec\u5206\u7c7b\u3001\u56de\u5f52\u3001\u805a\u7c7b\u6a21\u578b\u7b49\u4f20\u7edf\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\uff0c\u4e5f\u5305\u62ec\u6700\u8fd1\u5f00\u53d1\u548c\u5e7f\u6cdb\u4f7f\u7528\u7684\u6d89\u53ca\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\u7684\u6280\u672f\u3002\u672c\u6587\u8fd8\u8bb0\u5f55\u4e86\u4e00\u4e9b\u6700\u4f73\u505a\u6cd5\u4e0e\u5165\u95e8\u8981\u70b9\uff0c\u5e76\u5c55\u671b\u4e86\u673a\u5668\u5b66\u4e60\u5e94\u7528\u4e8e\u751f\u7269\u5b66\u7684\u4e00\u4e9b\u6700\u4ee4\u4eba\u5174\u594b\u7684\u524d\u666f\u3002<\/strong><\/p>\n


<\/p>\n

\u7814\u7a76\u9886\u57df\uff1a\u673a\u5668\u5b66\u4e60\uff0c\u4eba\u5de5\u795e\u7ecf\u7f51\u7edc\uff0c\u751f\u7269\u6570\u636e<\/strong><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/strong><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\n
\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\n
<\/strong><\/section>\n<\/section>\n
\n

Joe G. Greener, Shaun M. Kandathil\u7b49 <\/span>| \u4f5c\u8005<\/strong><\/span><\/strong><\/p>\n

\u8d75\u96e8\u4ead<\/span> <\/span>| \u8bd1\u8005<\/strong><\/span><\/strong><\/span><\/p>\n

\u9648\u65af\u4fe1<\/span> | \u5ba1\u6821<\/strong><\/span><\/strong><\/span><\/p>\n

\u9093\u4e00\u96ea<\/span> | \u7f16\u8f91<\/strong><\/span><\/p>\n<\/section>\n


<\/section>\n


<\/p>\n

<\/p>\n

\n
\n
\n

\u8bba\u6587\u9898\u76ee\uff1a<\/span><\/p>\n

A guide to machine learning for biologists<\/span><\/p>\n

\u8bba\u6587\u94fe\u63a5\uff1a<\/span><\/section>\n
https:\/\/www.nature.com\/articles\/s41580-021-00407-0<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n


<\/p>\n

\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n

\u76ee\u5f55<\/strong><\/p>\n

\n

<\/section>\n
\u5f15\u5165<\/span><\/strong><\/section>\n
\u5173\u952e\u6982\u5ff5<\/span><\/strong><\/section>\n
\u4f20\u7edf\u673a\u5668\u5b66\u4e60<\/span><\/strong><\/section>\n
\u4eba\u5de5\u795e\u7ecf\u7f51\u7edc<\/span><\/strong><\/section>\n
\u751f\u7269\u5e94\u7528\u7684\u6311\u6218<\/span><\/strong><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

<\/section>\n
\u4eba\u7c7b\u901a\u8fc7\u89c2\u5bdf\u6765\u7406\u89e3\u5468\u56f4\u7684\u4e16\u754c\uff0c\u5e76\u5b66\u4e60\u9884\u6d4b\u63a5\u4e0b\u6765\u4f1a\u53d1\u751f\u4ec0\u4e48\u3002\u8bbe\u60f3\u4e00\u4e2a\u5b69\u5b50\u5b66\u4e60\u63a5\u7403\uff1a\u8fd9\u4e2a\u5b69\u5b50\u901a\u5e38\u5bf9\u63a7\u5236\u629b\u7403\u8fd0\u52a8\u7684\u7269\u7406\u5b9a\u5f8b\u4e00\u65e0\u6240\u77e5\uff1b\u7136\u800c\uff0c\u901a\u8fc7\u89c2\u5bdf\u3001\u8bd5\u9519\u7684\u8fc7\u7a0b\uff0c\u5979\/\u4ed6\u4f1a\u8c03\u6574\u5bf9\u7403\u8fd0\u52a8\u7684\u7406\u89e3\uff0c\u4ee5\u53ca\u5982\u4f55\u79fb\u52a8\u8eab\u4f53\uff0c\u76f4\u5230\u80fd\u591f\u53ef\u9760\u5730\u63a5\u4f4f\u7403\u3002\u6362\u53e5\u8bdd\u8bf4\uff0c\u8fd9\u4e2a\u5b69\u5b50\u901a\u8fc7\u5efa\u7acb\u4e00\u4e2a\u8db3\u591f\u51c6\u786e\u548c\u6709\u7528\u7684\u8fc7\u7a0b\u201c\u6a21\u578b\u201d\uff0c\u5bf9\u6570\u636e\u53cd\u590d\u6d4b\u8bd5\u8fd9\u4e2a\u6a21\u578b\uff0c\u5e76\u5bf9\u6a21\u578b\u8fdb\u884c\u4fee\u6b63\u4f7f\u5176\u66f4\u597d\uff0c\u5b66\u4f1a\u4e86\u5982\u4f55\u63a5\u7403\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u201c\u673a\u5668\u5b66\u4e60\u201d\u6cdb\u6307\u7528\u9884\u6d4b\u6a21\u578b\u62df\u5408\u6570\u636e\u6216\u8bc6\u522b\u6570\u636e\u4e2d\u7684\u4fe1\u606f\u5206\u7ec4\u7684\u8fc7\u7a0b<\/span><\/strong>\u3002\u673a\u5668\u5b66\u4e60\u9886\u57df\u672c\u8d28\u4e0a\u8bd5\u56fe\u8fd1\u4f3c\u6216\u6a21\u4eff\u4eba\u7c7b\u8bc6\u522b\u6a21\u5f0f\u7684\u80fd\u529b\u2014\u2014\u5c3d\u7ba1\u662f\u4ee5\u201c\u8ba1\u7b97\u201d\u8fd9\u6837\u5ba2\u89c2\u7684\u65b9\u5f0f\u3002\u5f53\u4eba\u4eec\u60f3\u8981\u5206\u6790\u7684\u6570\u636e\u96c6\u56e0\u4e3a\u592a\u5927<\/span>\uff08\u8bb8\u591a\u5355\u72ec\u7684\u6570\u636e\u70b9\uff09<\/span>\u6216\u592a\u590d\u6742<\/span>\uff08\u5305\u542b\u5927\u91cf\u7279\u5f81\uff09<\/span>\u800c\u65e0\u6cd5\u8fdb\u884c\u4eba\u5de5\u5206\u6790\uff0c\u6216\u8005\u5f53\u4eba\u4eec\u9700\u8981\u81ea\u52a8\u5316\u6570\u636e\u5206\u6790\u7684\u8fc7\u7a0b\uff0c\u6765\u5efa\u7acb\u53ef\u91cd\u590d\u4e14\u7701\u65f6\u7684\u5de5\u4f5c\u6d41\u7a0b\u65f6\uff0c\u673a\u5668\u5b66\u4e60\u5c31\u5f88\u6709\u7528\u3002\u6765\u81ea\u751f\u7269\u5b9e\u9a8c\u7684\u6570\u636e\u901a\u5e38\u5c31\u662f\u8fd9\u6837\u7684\u3002\u5728\u8fc7\u53bb\u7684\u51e0\u5341\u5e74\u91cc\uff0c\u751f\u7269\u6570\u636e\u96c6\u7684\u89c4\u6a21\u4e0e\u590d\u6742\u6027\u90fd\u5927\u5e45\u589e\u957f\u3002\u56e0\u6b64\uff0c\u9664\u4e86\u638c\u63e1\u4e00\u4e9b\u53ef\u4ee5\u7528\u6765\u89e3\u91ca\u5927\u91cf\u6570\u636e\u7684\u5b9e\u7528\u65b9\u6cd5\uff0c\u900f\u5f7b\u7406\u89e3\u6240\u4f7f\u7528\u7684\u8fd9\u4e9b\u6280\u672f\u4e5f\u53d8\u5f97\u8d8a\u6765\u8d8a\u91cd\u8981\u3002\u673a\u5668\u5b66\u4e60\u5df2\u7ecf\u5728\u751f\u7269\u9886\u57df\u4e2d\u4f7f\u7528\u4e86\u51e0\u5341\u5e74\uff0c\u4f46\u5b83\u7684\u91cd\u8981\u6027\u7a33\u6b65\u589e\u957f\uff0c\u51e0\u4e4e\u5df2\u7ecf\u5e94\u7528\u4e8e\u751f\u7269\u5b66\u7684\u6bcf\u4e2a\u9886\u57df\u3002\u7136\u800c\uff0c\u76f4\u5230\u8fc7\u53bb\u51e0\u5e74\u4e2d\uff0c\u8be5\u9886\u57df\u624d\u5f00\u59cb\u5bf9\u53ef\u7528\u7b56\u7565\u8fdb\u884c\u66f4\u4e25\u683c\u7684\u5ba1\u89c6\uff0c\u5e76\u5f00\u59cb\u8bc4\u4f30\u5728\u4e0d\u540c\u7684\u573a\u666f\u4e0b\uff0c\u54ea\u4e9b\u65b9\u6cd5\u6700\u9002\u5408\uff0c\u6216\u8005\u6839\u672c\u4e0d\u5408\u9002\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u8fd9\u7bc7\u7efc\u8ff0\u65e8\u5728\u8ba9\u751f\u7269\u5b66\u5bb6\u4e86\u89e3\u5982\u4f55\u5f00\u59cb\u7406\u89e3\u548c\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u6280\u672f\u3002\u672c\u6587\u4e0d\u6253\u7b97\u5bf9\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u89e3\u51b3\u751f\u7269\u95ee\u9898\u7684\u6587\u7ae0\u8fdb\u884c\u5168\u9762\u7684\u6587\u732e\u7efc\u8ff0\uff0c\u6216\u63cf\u8ff0\u5404\u79cd\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u7684\u8be6\u7ec6\u6570\u5b66\u57fa\u7840 [2, 3]\u3002\u76f8\u53cd\uff0c\u672c\u6587\u4e13\u6ce8\u4e8e\u5c06\u7279\u5b9a\u6280\u672f\u4e0e\u4e0d\u540c\u7c7b\u578b\u7684\u751f\u7269\u6570\u636e\u8054\u7cfb\u8d77\u6765<\/span>\uff08\u7c7b\u4f3c\u7684\u7efc\u8ff0\u662f\u9488\u5bf9\u7279\u5b9a\u7684\u751f\u7269\u5b66\u79d1\u7684\uff1b\u8be6\u89c1\u53c2\u8003\u6587\u732e [4-11]\uff09<\/span>\u3002\u672c\u6587\u8fd8\u5c1d\u8bd5\u63d0\u70bc\u51fa\u4e00\u4e9b\u5173\u4e8e\u5982\u4f55\u5b9e\u9645\u8fdb\u884c\u8bad\u7ec3\u548c\u6539\u8fdb\u6a21\u578b\u8fc7\u7a0b\u7684\u6700\u4f73\u505a\u6cd5\u3002\u751f\u7269\u6570\u636e\u7684\u590d\u6742\u6027\u4e3a\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u6280\u672f\u8fdb\u884c\u5206\u6790\u5e26\u6765\u4e86\u9677\u9631\u548c\u673a\u4f1a\u3002\u4e3a\u4e86\u89e3\u51b3\u8fd9\u4e9b\u95ee\u9898\uff0c\u672c\u6587\u8ba8\u8bba\u4e86\u5f71\u54cd\u6210\u679c\u6709\u6548\u6027\u7684\u5e38\u89c1\u95ee\u9898\uff0c\u5e76\u63d0\u4f9b\u4e86\u5982\u4f55\u907f\u514d\u8fd9\u4e9b\u95ee\u9898\u7684\u6307\u5bfc\u3002\u8fd9\u7bc7\u7efc\u8ff0\u7684\u5927\u90e8\u5206\u5185\u5bb9\u90fd\u81f4\u529b\u4e8e\u63cf\u8ff0\u8bb8\u591a\u673a\u5668\u5b66\u4e60\u6280\u672f\uff0c\u5728\u6bcf\u79cd\u60c5\u51b5\u4e0b\uff0c\u672c\u6587\u90fd\u63d0\u4f9b\u4e86\u9002\u5f53\u5e94\u7528\u8be5\u65b9\u6cd5\u4ee5\u53ca\u5982\u4f55\u89e3\u91ca\u7ed3\u679c\u7684\u793a\u4f8b\u3002\u8ba8\u8bba\u7684\u65b9\u6cd5\u5305\u62ec\u4f20\u7edf\u7684\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u2014\u2014\u5b83\u4eec\u5728\u8bb8\u591a\u60c5\u51b5\u4e0b\u4ecd\u7136\u662f\u6700\u4f73\u9009\u62e9\u2014\u2014\u4ee5\u53ca\u57fa\u4e8e\u4eba\u5de5\u795e\u7ecf\u7f51\u7edc\u7684\u6df1\u5ea6\u5b66\u4e60\uff0c\u8fd9\u4e9b\u65b9\u6cd5\u6b63\u5728\u6210\u4e3a\u8bb8\u591a\u4efb\u52a1\u7684\u6700\u6709\u6548\u65b9\u6cd5\u3002\u672c\u6587\u6700\u540e\u63cf\u8ff0\u4e86\u5c06\u673a\u5668\u5b66\u4e60\u7eb3\u5165\u751f\u7269\u5b66\u6570\u636e\u5206\u6790\u901a\u7528\u6d41\u7a0b\u7684\u672a\u6765\u662f\u4ec0\u4e48\u6837\u7684\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u5728\u751f\u7269\u5b66\u4e2d\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u6709\u4e24\u4e2a\u76ee\u6807\u3002\u9996\u5148\u662f\u5728\u7f3a\u4e4f\u5b9e\u9a8c\u6570\u636e\u7684\u5730\u65b9\u505a\u51fa\u51c6\u786e\u7684\u9884\u6d4b\uff0c\u5e76\u5229\u7528\u8fd9\u4e9b\u9884\u6d4b\u6765\u6307\u5bfc\u672a\u6765\u7684\u7814\u7a76\u5de5\u4f5c<\/strong>\u3002\u7136\u800c\uff0c\u4f5c\u4e3a\u79d1\u5b66\u5bb6\uff0c\u6211\u4eec\u5bfb\u6c42\u4e86\u89e3\u4e16\u754c\uff0c\u56e0\u6b64\u7b2c\u4e8c\u4e2a\u76ee\u6807\u662f\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u6765\u8fdb\u4e00\u6b65\u4e86\u89e3\u751f\u7269\u5b66<\/strong>\u3002\u672c\u6307\u5357\u8ba8\u8bba\u4e86\u5728\u673a\u5668\u5b66\u4e60\u4e2d\uff0c\u8fd9\u4e24\u4e2a\u76ee\u6807\u5982\u4f55\u7ecf\u5e38\u53d1\u751f\u51b2\u7a81\uff0c\u4ee5\u53ca\u5982\u4f55\u4ece\u901a\u5e38\u88ab\u89c6\u4e3a\u201c\u9ed1\u7bb1\u201d\u7684\u6a21\u578b\u4e2d\u63d0\u53d6\u7406\u89e3\uff1a\u6bd5\u7adf\u5b83\u4eec\u7684\u5185\u90e8\u5de5\u4f5c\u539f\u7406\u96be\u4ee5\u7406\u89e3[12]\u3002<\/span><\/section>\n
 <\/span>
<\/section>\n


<\/p>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

\u5173\u952e\u6982\u5ff5<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n


<\/p>\n

\u672c\u6587\u9996\u5148\u4ecb\u7ecd\u673a\u5668\u5b66\u4e60\u4e2d\u7684\u4e00\u4e9b\u5173\u952e\u6982\u5ff5\u3002\u5728\u53ef\u80fd\u7684\u60c5\u51b5\u4e0b\uff0c\u672c\u6587\u4f1a\u7528\u751f\u7269\u5b66\u6587\u732e\u4e2d\u7684\u4f8b\u5b50\u6765\u8bf4\u660e\u8fd9\u4e9b\u6982\u5ff5\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u4e00\u822c\u672f\u8bed<\/span><\/strong><\/span>\u3002\u4e00\u4e2a\u6570\u636e\u96c6\u5305\u542b\u8bb8\u591a\u6570\u636e\u70b9<\/span>\uff08data points\uff09<\/span>\u6216\u5b9e\u4f8b<\/span>\uff08instances\uff09<\/span>\uff0c\u6bcf\u4e2a\u6570\u636e\u70b9\u6216\u5b9e\u4f8b\u90fd\u53ef\u4ee5\u88ab\u8ba4\u4e3a\u662f\u6765\u81ea\u5b9e\u9a8c\u7684\u5355\u4e2a\u89c2\u5bdf\u3002\u6bcf\u4e2a\u6570\u636e\u70b9\u90fd\u7531\u4e00\u5b9a\u6570\u91cf<\/span>\uff08\u901a\u5e38\u662f\u56fa\u5b9a\u7684\uff09<\/span>\u7684\u7279\u5f81<\/span>\uff08features\uff09<\/span>\u6240\u63cf\u8ff0\u3002\u6b64\u7c7b\u7279\u5f81\u7684\u4f8b\u5b50\u5305\u62ec\u957f\u5ea6\u3001\u65f6\u95f4\u3001\u6d53\u5ea6\u548c\u57fa\u56e0\u8868\u8fbe\u6c34\u5e73\u3002\u673a\u5668\u5b66\u4e60\u7684\u4efb\u52a1\u662f\u5bf9\u5e0c\u671b\u6a21\u578b\u8f93\u51fa\u7684\u503c\u8fdb\u884c\u5ba2\u89c2\u89c4\u8303\u3002\u4f8b\u5982\uff0c\u5bf9\u4e8e\u7814\u7a76\u57fa\u56e0\u968f\u65f6\u95f4\u8868\u8fbe\u7684\u5b9e\u9a8c\uff0c\u7814\u7a76\u4eba\u5458\u53ef\u80fd\u60f3\u8981\u9884\u6d4b\u7279\u5b9a\u4ee3\u8c22\u7269\u8f6c\u5316\u4e3a\u53e6\u4e00\u4e2a\u7269\u79cd\u7684\u901f\u7387\u3002\u5728\u8fd9\u79cd\u60c5\u51b5\u4e0b\uff0c\u7279\u5f81\u201c\u57fa\u56e0\u8868\u8fbe\u6c34\u5e73\u201d\u548c\u201c\u65f6\u95f4\u201d\u53ef\u4ee5\u79f0\u4e3a\u8f93\u5165\u7279\u5f81\u6216\u6a21\u578b\u7684\u7b80\u5355\u8f93\u5165\uff0c\u201c\u8f6c\u6362\u7387\u201d\u5c06\u662f\u6a21\u578b\u7684\u671f\u671b\u8f93\u51fa\uff1b\u4e5f\u5c31\u662f\u8bf4\uff0c\u7814\u7a76\u4eba\u5458\u6709\u5174\u8da3\u9884\u6d4b\u7684\u6570\u91cf\u3002\u4e00\u4e2a\u6a21\u578b\u53ef\u4ee5\u6709\u4efb\u610f\u6570\u91cf\u7684\u8f93\u5165\u548c\u8f93\u51fa\u7279\u5f81\u3002\u7279\u5f81\u53ef\u4ee5\u662f\u8fde\u7eed\u7684<\/span>\uff08\u53d6\u8fde\u7eed\u6570\u503c\uff09<\/span>\u6216\u5206\u7c7b\u7684<\/span>\uff08\u53ea\u53d6\u79bb\u6563\u503c\uff09<\/span>\u3002\u5f88\u591a\u65f6\u5019\uff0c\u5206\u7c7b\u7279\u5f81\u53ea\u662f\u4e8c\u5143\u7684\uff0c\u8981\u4e48\u662f\u771f (1)\uff0c\u8981\u4e48\u662f\u5047 (0)\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u76d1\u7763\u5b66\u4e60\u548c\u65e0\u76d1\u7763\u5b66\u4e60<\/span><\/strong><\/span>\u3002\u201c\u76d1\u7763\u673a\u5668\u5b66\u4e60\u201d<\/span>\uff08supervised learning\uff09<\/span>\u662f\u6307\u5c06\u6a21\u578b\u4e0e\u5df2\u6807\u8bb0\u7684\u6570\u636e<\/span>\uff08\u6216\u6570\u636e\u5b50\u96c6\uff09<\/span>\u8fdb\u884c\u62df\u5408\u2014\u2014\u5176\u4e2d\u5b58\u5728\u4e00\u4e9b\u5c5e\u6027\u7684\u771f\u5b9e\u503c<\/span>\uff08ground truth\uff09<\/span>\uff0c\u901a\u5e38\u901a\u8fc7\u5b9e\u9a8c\u6d4b\u91cf\u6216\u7531\u4eba\u7c7b\u5206\u914d\u3002\u4f8b\u5b50\u5305\u62ec\u5bf9\u86cb\u767d\u8d28\u4e8c\u7ea7\u7ed3\u6784\u7684\u9884\u6d4b [13] \u548c\u5bf9\u57fa\u56e0\u7ec4\u8c03\u63a7\u56e0\u5b50\u7684\u57fa\u56e0\u7ec4\u53ef\u53ca\u6027\u7684\u9884\u6d4b [14]\u3002\u5728\u8fd9\u4e24\u79cd\u60c5\u51b5\u4e0b\uff0c\u771f\u5b9e\u503c\u6700\u7ec8\u90fd\u6765\u81ea\u5b9e\u9a8c\u5ba4\u89c2\u5bdf\uff0c\u4f46\u8fd9\u4e9b\u539f\u59cb\u6570\u636e\u901a\u5e38\u4ee5\u67d0\u79cd\u65b9\u5f0f\u8fdb\u884c\u4e86\u9884\u5904\u7406\u3002\u4f8b\u5982\uff0c\u5728\u4e8c\u7ea7\u7ed3\u6784\u7684\u60c5\u51b5\u4e0b\uff0c\u771f\u5b9e\u503c\u6765\u81ea\u5206\u6790\u86cb\u767d\u8d28\u6570\u636e\u5e93\u4e2d\u7684\u86cb\u767d\u8d28\u6676\u4f53\u7ed3\u6784\u6570\u636e\uff0c\u5728\u540e\u4e00\u79cd\u60c5\u51b5\u4e0b\uff0c\u771f\u5b9e\u503c\u6765\u81eaDNA\u6d4b\u5e8f\u5b9e\u9a8c\u7684\u6570\u636e\u3002\u76f8\u6bd4\u4e4b\u4e0b\uff0c\u65e0\u76d1\u7763\u5b66\u4e60<\/span>\uff08unsupervised learning\uff09<\/span>\u65b9\u6cd5\u80fd\u591f\u8bc6\u522b\u672a\u6807\u8bb0\u6570\u636e\u4e2d\u7684\u6a21\u5f0f\uff0c\u800c\u65e0\u9700\u4ee5\u9884\u5b9a\u6807\u7b7e\u7684\u5f62\u5f0f\u5411\u673a\u5668\u5b66\u4e60\u7cfb\u7edf\u63d0\u4f9b\u771f\u5b9e\u503c\uff0c\u4f8b\u5982\u5728\u57fa\u56e0\u8868\u8fbe\u7814\u7a76\u4e2d\u627e\u5230\u5177\u6709\u76f8\u4f3c\u8868\u8fbe\u6c34\u5e73\u7684\u60a3\u8005\u5b50\u96c6[15] \u6216\u9884\u6d4b\u57fa\u56e0\u5e8f\u5217\u5171\u53d8\u5f02\u7684\u7a81\u53d8\u6548\u5e94 [16]\u3002\u6709\u65f6\u8fd9\u4e24\u79cd\u65b9\u6cd5\u5728\u534a\u76d1\u7763\u5b66\u4e60\u4e2d\u7ed3\u5408\uff0c\u5176\u4e2d\u5c11\u91cf\u6807\u8bb0\u6570\u636e\u4e0e\u5927\u91cf\u672a\u6807\u8bb0\u6570\u636e\u7ed3\u5408\u3002\u5728\u83b7\u53d6\u6807\u8bb0\u6570\u636e\u7684\u6210\u672c\u5f88\u9ad8\u7684\u60c5\u51b5\u4e0b\uff0c\u8fd9\u53ef\u4ee5\u63d0\u9ad8\u6027\u80fd\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u5206\u7c7b\u3001\u56de\u5f52\u548c\u805a\u7c7b\u95ee\u9898<\/span><\/strong><\/span>\u3002\u5f53\u95ee\u9898\u6d89\u53ca\u5c06\u6570\u636e\u70b9\u5206\u914d\u7ed9\u4e00\u7ec4\u79bb\u6563\u7c7b\u522b<\/span>\uff08\u4f8b\u5982\uff0c\u201c\u764c\u6027\u201d\u6216\u201c\u975e\u764c\u6027\u201d\uff09<\/span>\u65f6\uff0c\u8be5\u95ee\u9898\u79f0\u4e3a\u201c\u5206\u7c7b\u95ee\u9898\u201d<\/span>\uff08classification\uff09<\/span>\uff0c\u6267\u884c\u8fd9\u79cd\u5206\u7c7b\u7684\u4efb\u4f55\u7b97\u6cd5\u90fd\u53ef\u4ee5\u8bf4\u662f\u5206\u7c7b\u5668\u3002\u76f8\u6bd4\u4e4b\u4e0b\uff0c\u56de\u5f52<\/span>\uff08regression\uff09<\/span>\u6a21\u578b\u8f93\u51fa\u4e00\u7ec4\u8fde\u7eed\u7684\u503c\uff0c\u4f8b\u5982\u9884\u6d4b\u86cb\u767d\u8d28\u4e2d\u4e00\u4e2a\u6b8b\u57fa\u7a81\u53d8\u540e\u7684\u6298\u53e0\u81ea\u7531\u80fd\u53d8\u5316 [17]\u3002\u8fde\u7eed\u503c\u53ef\u4ee5\u9608\u503c\u5316\u6216\u4ee5\u5176\u4ed6\u65b9\u5f0f\u79bb\u6563\u5316\uff0c\u8fd9\u610f\u5473\u7740\u901a\u5e38\u53ef\u4ee5\u5c06\u56de\u5f52\u95ee\u9898\u91cd\u65b0\u8868\u8ff0\u4e3a\u5206\u7c7b\u95ee\u9898\u3002\u4f8b\u5982\uff0c\u4e0a\u8ff0\u81ea\u7531\u80fd\u53d8\u5316\u53ef\u4ee5\u5f52\u5165\u5bf9\u86cb\u767d\u8d28\u7a33\u5b9a\u6027\u6709\u5229\u6216\u4e0d\u5229\u7684\u503c\u8303\u56f4\u5185\u3002\u805a\u7c7b<\/span>\uff08clustering\uff09<\/span>\u65b9\u6cd5\u7528\u4e8e\u9884\u6d4b\u6570\u636e\u96c6\u4e2d\u76f8\u4f3c\u6570\u636e\u70b9\u7684\u5206\u7ec4\uff0c\u901a\u5e38\u57fa\u4e8e\u6570\u636e\u70b9\u4e4b\u95f4\u7684\u67d0\u79cd\u76f8\u4f3c\u6027\u5ea6\u91cf\u3002\u5b83\u4eec\u662f\u65e0\u76d1\u7763\u65b9\u6cd5\uff0c\u4e0d\u9700\u8981\u6570\u636e\u96c6\u4e2d\u7684\u793a\u4f8b\u5177\u6709\u6807\u7b7e\u3002\u4f8b\u5982\uff0c\u5728\u57fa\u56e0\u8868\u8fbe\u7814\u7a76\u4e2d\uff0c\u805a\u7c7b\u53ef\u4ee5\u627e\u5230\u5177\u6709\u76f8\u4f3c\u57fa\u56e0\u8868\u8fbe\u7684\u60a3\u8005\u5b50\u96c6\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u7c7b\u548c\u6807\u7b7e<\/span><\/strong><\/span>\u3002\u5206\u7c7b\u5668\u8fd4\u56de\u7684\u79bb\u6563\u503c\u53ef\u4ee5\u662f\u76f8\u4e92\u6392\u65a5\u7684\uff0c\u5728\u8fd9\u79cd\u60c5\u51b5\u4e0b\uff0c\u5b83\u4eec\u88ab\u79f0\u4e3a\u201c\u7c7b\u201d<\/span>\uff08class\uff09<\/span>\u3002\u5f53\u8fd9\u4e9b\u503c\u4e0d\u9700\u8981\u76f8\u4e92\u6392\u65a5\u65f6\uff0c\u5b83\u4eec\u88ab\u79f0\u4e3a\u201c\u6807\u7b7e\u201d<\/span>\uff08label\uff09<\/span>\u3002\u4f8b\u5982\uff0c\u86cb\u767d\u8d28\u7ed3\u6784\u4e2d\u7684\u6b8b\u57fa\u53ea\u80fd\u5c5e\u4e8e\u591a\u4e2a\u4e8c\u7ea7\u7ed3\u6784\u7c7b\u522b\u4e2d\u7684\u4e00\u4e2a\uff0c\u4f46\u53ef\u4ee5\u540c\u65f6\u3001\u4e0d\u4e92\u65a5\u5730\u88ab\u6807\u8bb0\u4e3a\u03b1-\u87ba\u65cb\u548c\u8de8\u819c\u3002\u7c7b\u548c\u6807\u7b7e\u901a\u5e38\u7531\u7f16\u7801\u8868\u793a<\/span>\uff08\u4f8b\u5982\uff0c\u72ec\u70ed\u7f16\u7801\uff0cone-hot encoding\uff09<\/span>\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u635f\u5931\u6216\u4ee3\u4ef7\u51fd\u6570<\/span><\/strong><\/span>\u3002\u673a\u5668\u5b66\u4e60\u6a21\u578b\u7684\u4e00\u4e2a\u6216\u591a\u4e2a\u8f93\u51fa\u4ece\u6765\u90fd\u4e0d\u662f\u7406\u60f3\u7684\uff0c\u4f1a\u504f\u79bb\u771f\u5b9e\u503c\u3002\u6d4b\u91cf\u8fd9\u79cd\u504f\u5dee\u7684\u6570\u5b66\u51fd\u6570\uff0c\u6216\u66f4\u7b3c\u7edf\u5730\u8bf4\uff0c\u6d4b\u91cf\u83b7\u5f97\u7684\u8f93\u51fa\u548c\u7406\u60f3\u7684\u8f93\u51fa\u4e4b\u95f4\u201c\u4e0d\u4e00\u81f4\u201d\u7684\u7a0b\u5ea6\u7684\u6570\u5b66\u51fd\u6570\uff0c\u88ab\u79f0\u4e3a\u201c\u635f\u5931\u51fd\u6570\u201d<\/span>\uff08loss function\uff09<\/span>\u6216\u201c\u4ee3\u4ef7\u51fd\u6570\u201d<\/span>\uff08cost function\uff09<\/span>\u3002\u5728\u76d1\u7763\u5b66\u4e60\u8bbe\u7f6e\u4e2d\uff0c\u635f\u5931\u51fd\u6570\u5c06\u8861\u91cf\u6a21\u578b\u7684\u8f93\u51fa\u76f8\u5bf9\u4e8e\u771f\u5b9e\u503c\u7684\u504f\u5dee\u3002\u4f8b\u5b50\u5305\u62ec\u56de\u5f52\u95ee\u9898\u7684\u5747\u65b9\u8bef\u5dee\u635f\u5931<\/span>\uff08mean squared error\uff09<\/span>\u548c\u5206\u7c7b\u95ee\u9898\u7684\u4e8c\u5143\u4ea4\u53c9\u71b5<\/span>\uff08binary cross entropy\uff09<\/span>\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u53c2\u6570\u548c\u8d85\u53c2\u6570<\/span><\/strong>\u3002\u6a21\u578b\u672c\u8d28\u4e0a\u662f\u5bf9\u4e00\u7ec4\u8f93\u5165\u7279\u5f81\u8fdb\u884c\u8fd0\u7b97\uff0c\u5e76\u4ea7\u751f\u4e00\u4e2a\u6216\u591a\u4e2a\u8f93\u51fa\u503c\u6216\u7279\u5f81\u7684\u6570\u5b66\u51fd\u6570\u3002\u4e3a\u4e86\u80fd\u591f\u5b66\u4e60\u8bad\u7ec3\u6570\u636e\uff0c\u6a21\u578b\u5305\u542b\u53ef\u8c03\u53c2\u6570<\/span>\uff08parameter\uff09<\/span>\uff0c\u5176\u503c\u53ef\u4ee5\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u6539\u53d8\uff0c\u4ee5\u5b9e\u73b0\u6a21\u578b\u7684\u6700\u4f73\u6027\u80fd<\/span>\uff08\u89c1\u4e0b\u6587\uff09<\/span>\u3002\u4f8b\u5982\uff0c\u5728\u4e00\u4e2a\u7b80\u5355\u7684\u56de\u5f52\u6a21\u578b\u4e2d\uff0c\u6bcf\u4e2a\u7279\u5f81\u90fd\u6709\u4e00\u4e2a\u4e58\u4ee5\u7279\u5f81\u503c\u7684\u53c2\u6570\uff0c\u7136\u540e\u5c06\u5b83\u4eec\u76f8\u52a0\u4ee5\u8fdb\u884c\u9884\u6d4b\u3002\u8d85\u53c2\u6570\u662f\u53ef\u8c03\u6574\u7684\u503c\uff0c\u4e0d\u88ab\u89c6\u4e3a\u6a21\u578b\u672c\u8eab\u7684\u4e00\u90e8\u5206\uff0c\u56e0\u4e3a\u5b83\u4eec\u5728\u8bad\u7ec3\u671f\u95f4\u4e0d\u4f1a\u66f4\u65b0\uff0c\u4f46\u4ecd\u7136\u5bf9\u6a21\u578b\u7684\u8bad\u7ec3\u53ca\u5176\u6027\u80fd\u4ea7\u751f\u5f71\u54cd\u3002\u8d85\u53c2\u6570\u7684\u4e00\u4e2a\u5e38\u89c1\u793a\u4f8b\u662f\u5b66\u4e60\u7387<\/span>\uff08learning rate\uff09<\/span>\uff0c\u5b83\u63a7\u5236\u6a21\u578b\u53c2\u6570\u5728\u8bad\u7ec3\u671f\u95f4\u66f4\u6539\u7684\u901f\u7387\u6216\u901f\u5ea6\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u8bad\u7ec3\u3001\u9a8c\u8bc1\u548c\u6d4b\u8bd5<\/span><\/strong>\u3002\u5728\u7528\u4e8e\u8fdb\u884c\u9884\u6d4b\u4e4b\u524d\uff0c\u6a21\u578b\u9700\u8981\u8fdb\u884c\u8bad\u7ec3<\/span>\uff08training\uff09<\/span>\uff0c\u8fd9\u6d89\u53ca\u81ea\u52a8\u8c03\u6574\u6a21\u578b\u7684\u53c2\u6570\u4ee5\u63d0\u9ad8\u5176\u6027\u80fd\u3002\u5728\u76d1\u7763\u5b66\u4e60\u8bbe\u7f6e\u4e2d\uff0c\u8fd9\u6d89\u53ca\u901a\u8fc7\u6700\u5c0f\u5316\u635f\u5931\u6216\u4ee3\u4ef7\u51fd\u6570\u7684\u5e73\u5747\u503c<\/span>\uff08\u524d\u9762\u5df2\u63cf\u8ff0\uff09<\/span>\u6765\u4fee\u6539\u53c2\u6570\uff0c\u4f7f\u6a21\u578b\u5728\u8bad\u7ec3\u6570\u636e\u96c6\u4e0a\u8868\u73b0\u826f\u597d\u3002\u901a\u5e38\uff0c\u5355\u72ec\u7684\u9a8c\u8bc1<\/span>\uff08validation\uff09<\/span>\u6570\u636e\u96c6\u7528\u4e8e\u76d1\u63a7\u4f46\u4e0d\u5f71\u54cd\u8bad\u7ec3\u8fc7\u7a0b\uff0c\u4ee5\u68c0\u6d4b\u6f5c\u5728\u7684\u8fc7\u5ea6\u62df\u5408<\/span>\uff08\u53c2\u89c1\u4e0b\u4e00\u8282\uff09<\/span>\u3002\u5728\u65e0\u76d1\u7763\u8bbe\u7f6e\u4e2d\uff0c\u4ee3\u4ef7\u51fd\u6570\u4ecd\u7136\u88ab\u6700\u5c0f\u5316\uff0c\u5c3d\u7ba1\u5b83\u4e0d\u518d\u6d89\u53ca\u771f\u5b9e\u503c\u3002\u4e00\u65e6\u6a21\u578b\u7ecf\u8fc7\u8bad\u7ec3\uff0c\u5c31\u53ef\u4ee5\u5728\u672a\u7528\u4e8e\u8bad\u7ec3\u7684\u6570\u636e\u4e0a\u8fdb\u884c\u6d4b\u8bd5<\/span>\uff08testing\uff09<\/span>\u3002\u6709\u5173\u6574\u4e2a\u8bad\u7ec3\u8fc7\u7a0b\u4ee5\u53ca\u5982\u4f55\u5728\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u4e4b\u95f4\u9002\u5f53\u62c6\u5206\u6570\u636e\u7684\u6307\u5357\uff0c\u8bf7\u53c2\u89c1\u65b9\u68461\u3002\u56fe1\u5c55\u793a\u4e86\u6574\u4e2a\u673a\u5668\u5b66\u4e60\u8fc7\u7a0b\u7684\u6d41\u7a0b\u56fe\u3002\u56fe2\u5c55\u793a\u4e86\u6a21\u578b\u8bad\u7ec3\u4e2d\u7684\u4e00\u4e9b\u6982\u5ff5\u3002<\/span><\/section>\n
   <\/span><\/section>\n
\u8fc7\u62df\u5408\u548c\u6b20\u62df\u5408<\/span><\/strong>\u3002\u5c06\u6a21\u578b\u62df\u5408\u5230\u8bad\u7ec3\u6570\u636e\u7684\u76ee\u7684\u662f\u6355\u6349\u6570\u636e\u4e2d\u53d8\u91cf\u4e4b\u95f4\u7684\u201c\u771f\u5b9e\u201d\u5173\u7cfb\uff0c\u4ee5\u4fbf\u6a21\u578b\u5bf9\u672c\u6b21\u672a\u5305\u542b\u7684<\/span>\uff08\u5373\u975e\u8bad\u7ec3\uff09<\/span>\u6570\u636e\u5177\u6709\u9884\u6d4b\u80fd\u529b\u3002\u8fc7\u62df\u5408\u6216\u6b20\u62df\u5408\u7684\u6a21\u578b\u5c06\u5bf9\u4e0d\u5728\u8bad\u7ec3\u96c6\u4e2d\u7684\u6570\u636e\u4ea7\u751f\u8f83\u5dee\u7684\u9884\u6d4b<\/span>\uff08\u56fe 2d\uff09<\/span>\u3002\u8fc7\u62df\u5408\u7684\u6a21\u578b\u4f1a\u5728\u8bad\u7ec3\u96c6\u4e2d\u7684\u6570\u636e\u4e0a\u4ea7\u751f\u5f88\u597d\u7684\u7ed3\u679c<\/span>\uff08\u901a\u5e38\u662f\u56e0\u4e3a\u53c2\u6570\u592a\u591a\uff09<\/span>\uff0c\u4f46\u5728\u771f\u5b9e\u6570\u636e\u4e0a\u4f1a\u4ea7\u751f\u5f88\u5dee\u7684\u7ed3\u679c\u3002\u56fe2d\u4e2d\u7684\u8fc7\u62df\u5408\u6a21\u578b\u6b63\u597d\u901a\u8fc7\u6bcf\u4e2a\u8bad\u7ec3\u70b9\uff0c\u56e0\u6b64\u5b83\u5728\u8bad\u7ec3\u96c6\u4e0a\u7684\u9884\u6d4b\u8bef\u5dee\u4e3a\u96f6\u3002\u7136\u800c\uff0c\u5f88\u660e\u663e\uff0c\u8fd9\u4e2a\u6a21\u578b\u5df2\u7ecf\u201c\u8bb0\u4f4f\u201d\u4e86\u8bad\u7ec3\u6570\u636e\uff0c\u4e0d\u592a\u53ef\u80fd\u5728\u771f\u5b9e\u6570\u636e\u4e0a\u4ea7\u751f\u597d\u7684\u7ed3\u679c\u3002\u76f8\u6bd4\u4e4b\u4e0b\uff0c\u6b20\u62df\u5408\u7684\u6a21\u578b\u65e0\u6cd5\u5145\u5206\u6355\u6349\u6570\u636e\u4e2d\u53d8\u91cf\u4e4b\u95f4\u7684\u5173\u7cfb\u3002\u8fd9\u53ef\u80fd\u662f\u7531\u4e8e\u6a21\u578b\u7c7b\u578b\u9009\u62e9\u4e0d\u6b63\u786e\u3001\u5bf9\u6570\u636e\u7684\u5047\u8bbe\u4e0d\u5b8c\u6574\u6216\u4e0d\u6b63\u786e\u3001\u6a21\u578b\u4e2d\u7684\u53c2\u6570\u592a\u5c11\u548c\/\u6216\u8bad\u7ec3\u8fc7\u7a0b\u4e0d\u5b8c\u6574\u3002\u5982\u56fe2d \u6240\u793a\u7684\u6b20\u62df\u5408\u6a21\u578b\u5bf9\u4e8e\u5b83\u8bd5\u56fe\u62df\u5408\u7684\u6570\u636e\u6765\u8bf4\u662f\u4e0d\u591f\u7684\uff1b\u5728\u8fd9\u79cd\u60c5\u51b5\u4e0b\uff0c\u5f88\u660e\u663e\u53d8\u91cf\u5177\u6709\u975e\u7ebf\u6027\u5173\u7cfb\uff0c\u4e0d\u80fd\u7528\u7b80\u5355\u7684\u7ebf\u6027\u6a21\u578b\u5145\u5206\u63cf\u8ff0\u3002\u56e0\u6b64\uff0c\u975e\u7ebf\u6027\u6a21\u578b\u66f4\u5408\u9002\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u5f52\u7eb3\u504f\u5dee\u548c\u504f\u5dee-\u65b9\u5dee\u6743\u8861<\/span><\/strong>\u3002\u6a21\u578b\u7684\u201c\u5f52\u7eb3\u504f\u5dee\u201d\u662f\u6307\u5728\u5b66\u4e60\u7b97\u6cd5\u4e2d\u505a\u51fa\u7684\u4e00\u7ec4\u5047\u8bbe\uff0c\u8fd9\u4e9b\u5047\u8bbe\u5bfc\u81f4\u5b83\u504f\u5411\u4e8e\u5b66\u4e60\u95ee\u9898\u7684\u7279\u5b9a\u89e3\u51b3\u65b9\u6848\uff0c\u800c\u4e0d\u662f\u5176\u4ed6\u89e3\u51b3\u65b9\u6848\u3002\u5b83\u53ef\u4ee5\u88ab\u8ba4\u4e3a\u662f\u6a21\u578b\u5bf9\u5b66\u4e60\u95ee\u9898\u7684\u7279\u5b9a\u7c7b\u578b\u89e3\u51b3\u65b9\u6848\u7684\u504f\u597d\u3002\u8fd9\u79cd\u504f\u597d\u901a\u5e38\u4f7f\u7528\u5176\u7279\u5b9a\u7684\u6570\u5b66\u5f62\u5f0f\u548c\/\u6216\u4f7f\u7528\u7279\u5b9a\u7684\u635f\u5931\u51fd\u6570\u88ab\u7f16\u7a0b\u5230\u6a21\u578b\u4e2d\u3002\u4f8b\u5982\uff0c\u5faa\u73af\u795e\u7ecf\u7f51\u7edc<\/span>\uff08recurrent neural networks\uff0cRNN\uff1b\u4e0b\u6587\u4e2d\u5c06\u8fdb\u884c\u8bb2\u89e3\uff09<\/span>\u7684\u5f52\u7eb3\u504f\u5dee\u662f\u8f93\u5165\u6570\u636e\u4e2d\u5b58\u5728\u987a\u5e8f\u4f9d\u8d56\u6027\uff0c\u4f8b\u5982\u4ee3\u8c22\u7269\u968f\u65f6\u95f4\u7684\u6d53\u5ea6\u3002\u8fd9\u79cd\u4f9d\u8d56\u5173\u7cfb\u5728 RNN \u7684\u6570\u5b66\u5f62\u5f0f\u4e2d\u5f97\u5230\u4e86\u660e\u786e\u7684\u89e3\u91ca\u3002\u4e0d\u540c\u6a21\u578b\u7c7b\u578b\u4e2d\u7684\u4e0d\u540c\u5f52\u7eb3\u504f\u5dee\u4f7f\u5b83\u4eec\u66f4\u9002\u5408\u5e76\u4e14\u901a\u5e38\u5bf9\u7279\u5b9a\u7c7b\u578b\u7684\u6570\u636e\u8868\u73b0\u66f4\u597d\u3002\u53e6\u4e00\u4e2a\u91cd\u8981\u6982\u5ff5\u662f\u504f\u5dee<\/span>\uff08bias\uff09<\/span>\u548c\u65b9\u5dee<\/span>\uff08variance\uff09<\/span>\u4e4b\u95f4\u7684\u6743\u8861\u3002\u53ef\u4ee5\u8bf4\u5177\u6709\u9ad8\u504f\u5dee\u7684\u6a21\u578b\u5bf9\u8bad\u7ec3\u6a21\u578b\u5177\u6709\u66f4\u5f3a\u7684\u7ea6\u675f\uff0c\u800c\u5177\u6709\u4f4e\u504f\u5dee\u7684\u6a21\u578b\u5bf9\u88ab\u5efa\u6a21\u7684\u5c5e\u6027\u505a\u51fa\u8f83\u5c11\u7684\u5047\u8bbe\uff0c\u5e76\u4e14\u7406\u8bba\u4e0a\u53ef\u4ee5\u5bf9\u591a\u79cd\u51fd\u6570\u7c7b\u578b\u8fdb\u884c\u5efa\u6a21\u3002\u6a21\u578b\u7684\u65b9\u5dee\u63cf\u8ff0\u4e86\u5728\u4e0d\u540c\u8bad\u7ec3\u6570\u636e\u96c6\u4e0a\u8fdb\u884c\u8bad\u7ec3\uff0c\u7ecf\u8fc7\u8bad\u7ec3\u7684\u6a21\u578b\u53d1\u751f\u7684\u53d8\u5316\u7684\u5927\u5c0f\u3002\u4e00\u822c\u6765\u8bf4\uff0c\u7814\u7a76\u4eba\u5458\u5e0c\u671b\u6a21\u578b\u5177\u6709\u975e\u5e38\u4f4e\u7684\u504f\u5dee\u548c\u4f4e\u65b9\u5dee\uff0c\u5c3d\u7ba1\u8fd9\u4e9b\u76ee\u6807\u7ecf\u5e38\u53d1\u751f\u51b2\u7a81\uff0c\u56e0\u4e3a\u5177\u6709\u4f4e\u504f\u5dee\u7684\u6a21\u578b\u901a\u5e38\u4f1a\u5728\u4e0d\u540c\u7684\u8bad\u7ec3\u96c6\u4e0a\u5b66\u4e60\u4e0d\u540c\u7684\u4fe1\u53f7\u3002\u63a7\u5236\u504f\u5dee-\u65b9\u5dee\u6743\u8861\u662f\u907f\u514d\u8fc7\u62df\u5408\u6216\u6b20\u62df\u5408\u7684\u5173\u952e\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\n
\n
\n

<\/section>\n

<\/section>\n<\/section>\n
\n
\n

\u65b9\u68461 | \u505a\u673a\u5668\u5b66\u4e60<\/span><\/strong><\/span><\/p>\n


<\/section>\n

\u8fd9\u4e2a\u65b9\u6846\u6982\u8ff0\u4e86\u5728\u8bad\u7ec3\u673a\u5668\u5b66\u4e60\u6a21\u578b\u65f6\u5e94\u8be5\u91c7\u53d6\u7684\u6b65\u9aa4\u3002\u4ee4\u4eba\u60ca\u8bb6\u7684\u662f\uff0c\u5173\u4e8e\u6a21\u578b\u9009\u62e9\u548c\u8bad\u7ec3\u8fc7\u7a0b\u7684\u6307\u5bfc\u5f88\u5c11 [146,147]\uff0c\u5bf9\u57ab\u811a\u77f3\u548c\u5931\u8d25\u6a21\u578b\u7684\u63cf\u8ff0\u4e5f\u5f88\u5c11\u88ab\u7eb3\u5165\u5df2\u53d1\u8868\u7684\u7814\u7a76\u6587\u7ae0\u3002\u5728\u63a5\u89e6\u4efb\u4f55\u673a\u5668\u5b66\u4e60\u4ee3\u7801\u4e4b\u524d\uff0c\u7b2c\u4e00\u6b65\u5e94\u8be5\u662f\u5145\u5206\u7406\u89e3\u624b\u5934\u7684\u6570\u636e\uff08\u8f93\u5165\uff09\u548c\u9884\u6d4b\u4efb\u52a1\uff08\u8f93\u51fa\uff09<\/strong>\u3002\u8fd9\u610f\u5473\u7740\u5bf9\u95ee\u9898\u7684\u751f\u7269\u5b66\u7406\u89e3\uff1a\u4f8b\u5982\u4e86\u89e3\u6570\u636e\u7684\u6765\u6e90\u548c\u566a\u58f0\u6e90\uff0c\u5e76\u77e5\u9053\uff0c\u6839\u636e\u751f\u7269\u5b66\u539f\u7406\uff0c\u8f93\u51fa\u7406\u8bba\u4e0a\u662f\u600e\u6837\u4ece\u8f93\u5165\u4e2d\u9884\u6d4b\u7684\u3002\u4f8b\u5982\uff0c\u53ef\u4ee5\u63a8\u65ad\u4e0d\u540c\u7684\u6c28\u57fa\u9178\u53ef\u80fd\u5bf9\u86cb\u767d\u8d28\u4e2d\u7684\u7279\u5b9a\u4e8c\u7ea7\u7ed3\u6784\u6709\u504f\u597d\uff0c\u56e0\u6b64\uff0c\u4ece\u86cb\u767d\u8d28\u5e8f\u5217\u4e2d\u6bcf\u4e2a\u4f4d\u7f6e\u7684\u6c28\u57fa\u9178\u9891\u7387\u9884\u6d4b\u4e8c\u7ea7\u7ed3\u6784\u662f\u6709\u610f\u4e49\u7684\u3002\u4e86\u89e3\u8f93\u5165\u548c\u8f93\u51fa\u5728\u8ba1\u7b97\u4e0a\u662f\u5982\u4f55\u5b58\u50a8\u7684\u4e5f\u5f88\u91cd\u8981\u3002\u5b83\u4eec\u662f\u5426\u5df2\u88ab\u5f52\u4e00\u5316\u4ee5\u9632\u6b62\u4e00\u4e2a\u7279\u5f81\u5bf9\u9884\u6d4b\u4ea7\u751f\u8fc7\u5927\u7684\u5f71\u54cd\uff1f\u5b83\u4eec\u662f\u7f16\u7801\u4e3a\u4e8c\u8fdb\u5236\u53d8\u91cf\u8fd8\u662f\u8fde\u7eed\u7f16\u7801\uff1f\u662f\u5426\u6709\u91cd\u590d\u7684\u6761\u76ee\uff1f\u662f\u5426\u7f3a\u5c11\u6570\u636e\u5143\u7d20\uff1f<\/span><\/p>\n


<\/span><\/p>\n

\u63a5\u4e0b\u6765\uff0c\u5e94\u62c6\u5206\u6570\u636e\u4ee5\u8fdb\u884c\u8bad\u7ec3\u3001\u9a8c\u8bc1\u548c\u6d4b\u8bd5\u3002\u6709\u591a\u79cd\u65b9\u6cd5\u53ef\u4ee5\u505a\u5230\u8fd9\u4e00\u70b9\uff0c\u56fe 2 \u4e2d\u663e\u793a\u4e86\u5176\u4e2d\u7684\u4e24\u79cd\u65b9\u6cd5\u30022a. \u8bad\u7ec3\u96c6\u7528\u4e8e\u76f4\u63a5\u66f4\u65b0\u88ab\u8bad\u7ec3\u6a21\u578b\u7684\u53c2\u6570\u3002\u9a8c\u8bc1\u96c6\u901a\u5e38\u5360\u53ef\u7528\u6570\u636e\u7684 10% \u5de6\u53f3\uff0c\u7528\u4e8e\u76d1\u63a7\u8bad\u7ec3\u3001\u9009\u62e9\u8d85\u53c2\u6570\u5e76\u9632\u6b62\u6a21\u578b\u8fc7\u5ea6\u62df\u5408\u8bad\u7ec3\u6570\u636e\u3002\u901a\u5e38\u4f7f\u7528k\u6298\u4ea4\u53c9\u9a8c\u8bc1\uff08k-fold cross validation\uff09\uff1a\u5c06\u8bad\u7ec3\u96c6\u5206\u6210 k \u4e2a\u5927\u5c0f\u5747\u5300\u7684\u5206\u533a\uff08\u4f8b\u5982\uff0c\u4e94\u4e2a\u6216\u5341\u4e2a\uff09\u4ee5\u5f62\u6210 k \u4e2a\u4e0d\u540c\u7684\u8bad\u7ec3\u548c\u9a8c\u8bc1\u96c6\uff0c\u5e76\u5728\u6bcf\u4e2a\u5206\u533a\u4e4b\u95f4\u6bd4\u8f83\u6027\u80fd\u4ee5\u9009\u62e9\u6700\u4f73\u8d85\u53c2\u6570\u3002\u6d4b\u8bd5\u96c6\uff0c\u6709\u65f6\u4e5f\u79f0\u4e3a\u201c\u7559\u51fa\u96c6\u201d\uff0c\u901a\u5e38\u4e5f\u5360\u53ef\u7528\u6570\u636e\u7684 10% \u5de6\u53f3\uff0c\u7528\u4e8e\u8bc4\u4f30\u6a21\u578b\u5728\u672a\u7528\u4e8e\u8bad\u7ec3\u6216\u9a8c\u8bc1\u7684\u6570\u636e\u4e0a\u7684\u6027\u80fd\uff08\u5373\u4f30\u8ba1\u5176\u9884\u671f\u7684\u771f\u5b9e\u8868\u73b0\uff09\u3002\u6d4b\u8bd5\u96c6\u5e94\u8be5\u53ea\u5728\u7814\u7a76\u7684\u6700\u540e\u4f7f\u7528\u4e00\u6b21\uff0c\u6216\u8005\u5c3d\u53ef\u80fd\u4e0d\u9891\u7e41\u5730\u4f7f\u7528 [27, 38] \uff0c\u4ee5\u907f\u514d\u6a21\u578b\u62df\u5408\u4e86\u6d4b\u8bd5\u96c6\u3002\u5173\u4e8e\u5236\u4f5c\u4e00\u4e2a\u516c\u5e73\u7684\u6d4b\u8bd5\u96c6\u65f6\u9700\u8981\u8003\u8651\u7684\u95ee\u9898\uff0c\u8bf7\u53c2\u89c1\u6570\u636e\u6cc4\u6f0f\u4e00\u8282\u3002<\/span><\/p>\n


<\/span><\/p>\n

\u4e0b\u4e00\u6b65\u662f\u6a21\u578b\u9009\u62e9\uff0c\u8fd9\u53d6\u51b3\u4e8e\u6570\u636e\u7684\u6027\u8d28\u548c\u9884\u6d4b\u4efb\u52a1\uff0c\u603b\u7ed3\u5728\u56fe1\u4e2d\u30021. \u8bad\u7ec3\u96c6\u7528\u4e8e\u8bad\u7ec3\u6a21\u578b\uff0c\u6309\u7167\u6240\u7528\u8f6f\u4ef6\u6846\u67b6\u7684\u6700\u4f73\u505a\u6cd5\u3002\u4e3a\u8fbe\u5230\u6700\u4f73\u6027\u80fd\uff0c\u5927\u591a\u6570\u65b9\u6cd5\u90fd\u6709\u4e00\u4e9b\u8d85\u53c2\u6570\u9700\u8981\u8c03\u6574\u3002\u8fd9\u53ef\u4ee5\u4f7f\u7528\u968f\u673a\u641c\u7d22\u6216\u7f51\u683c\u641c\u7d22\u6765\u5b8c\u6210\uff0c\u5e76\u4e14\u53ef\u4ee5\u4e0e\u4e0a\u9762\u6982\u8ff0\u7684k\u6298\u4ea4\u53c9\u9a8c\u8bc1\u76f8\u7ed3\u5408 [27]\u3002\u7814\u7a76\u4eba\u5458\u5e94\u8be5\u8003\u8651\u6a21\u578b\u96c6\u6210\uff0c\u5c06\u591a\u4e2a\u76f8\u4f3c\u6a21\u578b\u7684\u8f93\u51fa\u7b80\u5355\u5e73\u5747\uff0c\u4ee5\u63d0\u4f9b\u4e00\u79cd\u76f8\u5bf9\u53ef\u9760\u7684\u65b9\u6cd5\uff0c\u6765\u63d0\u9ad8\u5efa\u6a21\u4efb\u52a1\u7684\u6574\u4f53\u51c6\u786e\u6027\u3002\u6700\u540e\uff0c\u5e94\u8be5\u8bc4\u4f30\u6a21\u578b\u5728\u6d4b\u8bd5\u96c6\uff08\u89c1\u4e0a\u6587\uff09\u4e0a\u7684\u51c6\u786e\u6027\u3002<\/span><\/p>\n<\/section>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n
 <\/span><\/section>\n
<\/section>\n

\u56fe1. \u9009\u62e9\u548c\u8bad\u7ec3\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u3002\u8bad\u7ec3\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u7684\u6574\u4e2a\u8fc7\u7a0b\u663e\u793a\u5728\u4e0a\u90e8\u3002\u4e0b\u9762\u7ed9\u51fa\u4e86\u5e2e\u52a9\u7814\u7a76\u4eba\u5458\u9009\u62e9\u6a21\u578b\u7684\u51b3\u7b56\u6811\u3002\u8fd9\u4e2a\u6d41\u7a0b\u56fe\u65e8\u5728\u4f5c\u4e3a\u4e00\u4e2a\u89c6\u89c9\u6307\u5357\uff0c\u5c06\u672c\u7efc\u8ff0\u4e2d\u6982\u8ff0\u7684\u6982\u5ff5\u8054\u7cfb\u8d77\u6765\u3002\u7136\u800c\uff0c\u50cf\u8fd9\u6837\u7684\u7b80\u5355\u6982\u8ff0\u4e0d\u80fd\u6db5\u76d6\u6240\u6709\u60c5\u51b5\u3002\u4f8b\u5982\uff0c\u673a\u5668\u5b66\u4e60\u8981\u53d8\u5f97\u9002\u7528\u6240\u9700\u7684\u6570\u636e\u70b9\u6570\u91cf\uff0c\u53d6\u51b3\u4e8e\u6bcf\u4e2a\u6570\u636e\u70b9\u53ef\u7528\u7684\u7279\u5f81\u6570\u91cf\u2014\u2014\u7279\u5f81\u8d8a\u591a\uff0c\u9700\u8981\u7684\u6570\u636e\u70b9\u8d8a\u591a\uff0c\u8fd8\u53d6\u51b3\u4e8e\u6240\u4f7f\u7528\u7684\u6a21\u578b\u3002\u8fd8\u6709\u4e00\u4e9b\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u53ef\u4ee5\u5904\u7406\u672a\u6807\u8bb0\u7684\u6570\u636e\u3002<\/span>
<\/p>\n

 <\/span><\/section>\n
<\/section>\n

\u56fe2. \u8bad\u7ec3\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u3002\uff08a\uff09<\/strong>\u53ef\u7528\u6570\u636e\u901a\u5e38\u5206\u4e3a\u8bad\u7ec3\u3001\u9a8c\u8bc1\u548c\u6d4b\u8bd5\u96c6\u3002\u8bad\u7ec3\u96c6\u76f4\u63a5\u7528\u4e8e\u8bad\u7ec3\u6a21\u578b\uff0c\u9a8c\u8bc1\u96c6\u7528\u4e8e\u76d1\u63a7\u8bad\u7ec3\uff0c\u6d4b\u8bd5\u96c6\u7528\u4e8e\u8bc4\u4f30\u6a21\u578b\u7684\u6027\u80fd\u3002\u4e5f\u53ef\u4ee5\u4f7f\u7528\u5e26\u6709\u6d4b\u8bd5\u96c6\u7684k\u6298\u4ea4\u53c9\u9a8c\u8bc1\u3002\uff08b\uff09<\/strong> \u72ec\u70ed\u7f16\u7801\u662f\u8868\u793a\u5206\u7c7b\u8f93\u5165\u7684\u5e38\u7528\u65b9\u6cd5\uff0c\u53ea\u5141\u8bb8\u4ece\u591a\u79cd\u53ef\u80fd\u6027\uff08\u5728\u8fd9\u91cc\u662f\u4e09\u79cd\u53ef\u80fd\u7684\u86cb\u767d\u8d28\u4e8c\u7ea7\u7ed3\u6784\u7c7b\u522b\uff09\u4e2d\u8fdb\u884c\u5355\u4e00\u9009\u62e9\u3002\u7f16\u7801\u7684\u7ed3\u679c\u662f\u4e00\u4e2a\u5305\u542b\u4e09\u4e2a\u6570\u5b57\u7684\u5411\u91cf\uff0c\u9664\u4e86\u88ab\u5360\u7528\u7684\u7c7b\u8bbe\u7f6e\u4e3a 1 \u5916\uff0c\u6240\u6709\u6570\u5b57\u90fd\u7b49\u4e8e 0\u3002\u8fd9\u4e2a\u5411\u91cf\u88ab\u673a\u5668\u5b66\u4e60\u6a21\u578b\u4f7f\u7528\u3002\uff08c\uff09<\/strong>\u8fde\u7eed\u7f16\u7801\u8868\u793a\u6570\u5b57\u8f93\u5165\uff0c\u5728\u8fd9\u79cd\u60c5\u51b5\u4e0b\u662f\u56fe\u50cf\u4e2d\u50cf\u7d20\u7684\u7ea2\u8272\u3001\u7eff\u8272\u548c\u84dd\u8272 (RGB) \u503c\u3002\u7ed3\u679c\u540c\u6837\u662f\u4e00\u4e2a\u5305\u542b\u4e09\u4e2a\u6570\u5b57\u7684\u5411\u91cf\uff0c\u5bf9\u5e94\u4e8e\u50cf\u7d20\u4e2d\u7ea2\u8272\u3001\u7eff\u8272\u548c\u84dd\u8272\u7684\u6570\u91cf\u3002\uff08d\uff09<\/strong>\u672a\u80fd\u5b66\u4e60\u5230\u53d8\u91cf\u4e4b\u95f4\u7684\u6f5c\u5728\u5173\u7cfb\u79f0\u4e3a\u201c\u6b20\u62df\u5408\u201d\uff0c\u800c\u5b66\u4e60\u4e86\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u566a\u58f0\u79f0\u4e3a\u201c\u8fc7\u62df\u5408\u201d\u3002\u6b20\u62df\u5408\u53ef\u80fd\u662f\u7531\u4e8e\u4f7f\u7528\u4e86\u4e0d\u591f\u590d\u6742\u7684\u6a21\u578b\u6765\u63cf\u8ff0\u4fe1\u53f7\u3002\u800c\u8fc7\u62df\u5408\u53ef\u80fd\u662f\u7531\u4e8e\u4f7f\u7528\u4e86\u53c2\u6570\u8fc7\u591a\u7684\u6a21\u578b\u6216\u5728\u5b66\u4e60\u4e86\u53d8\u91cf\u4e4b\u95f4\u7684\u771f\u5b9e\u5173\u7cfb\u540e\u7ee7\u7eed\u8bad\u7ec3\u3002\uff08e\uff09<\/strong>\u6a21\u578b\u7684\u5b66\u4e60\u7387\u51b3\u5b9a\u4e86\u5728\u8bad\u7ec3\u795e\u7ecf\u7f51\u7edc\u6216\u4e00\u4e9b\u4f20\u7edf\u6a21\u578b\uff08\u5982\u68af\u5ea6\u63d0\u5347\uff09\u65f6\u8c03\u6574\u5b66\u4e60\u53c2\u6570\u7684\u901f\u5ea6\u3002\u4f4e\u5b66\u4e60\u7387\u4f1a\u5bfc\u81f4\u8bad\u7ec3\u7f13\u6162\uff0c\u8fd9\u65e2\u8017\u65f6\u53c8\u9700\u8981\u76f8\u5f53\u5927\u7684\u8ba1\u7b97\u80fd\u529b\u3002\u76f8\u53cd\uff0c\u9ad8\u5b66\u4e60\u7387\u4f1a\u5bfc\u81f4\u5feb\u901f\u6536\u655b\u5230\u4e00\u4e2a\u975e\u6700\u4f18\u89e3\u548c\u6a21\u578b\u6027\u80fd\u4e0d\u4f73\u3002\uff08f\uff09<\/strong>\u63d0\u524d\u505c\u6b62\uff08early stopping\uff09\u662f\u5728\u9a8c\u8bc1\u96c6\u4e0a\u7684\u635f\u5931\u51fd\u6570\u5f00\u59cb\u589e\u52a0\u65f6\u7ec8\u6b62\u8bad\u7ec3\u7684\u8fc7\u7a0b\uff0c\u5373\u4f7f\u8bad\u7ec3\u96c6\u4e0a\u7684\u635f\u5931\u51fd\u6570\u4ecd\u5728\u51cf\u5c11\u3002\u4f7f\u7528\u63d0\u524d\u505c\u6b62\u53ef\u4ee5\u9632\u6b62\u8fc7\u62df\u5408\u3002<\/span><\/p>\n


<\/section>\n


<\/p>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

\u4f20\u7edf\u673a\u5668\u5b66\u4e60<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n


<\/p>\n

\u672c\u6587\u5c06\u8ba8\u8bba\u51e0\u79cd\u5173\u952e\u7684\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\uff0c\u91cd\u70b9\u4ecb\u7ecd\u5b83\u4eec\u7684\u7279\u5b9a\u4f18\u52bf\u548c\u52a3\u52bf\u3002\u88681\u663e\u793a\u4e86\u4e0d\u540c\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u7684\u6bd4\u8f83\u3002\u672c\u8282\u5c06\u9996\u5148\u8ba8\u8bba\u4e0d\u57fa\u4e8e\u795e\u7ecf\u7f51\u7edc\u7684\u65b9\u6cd5\uff0c\u6709\u65f6\u79f0\u4e3a\u201c\u4f20\u7edf\u673a\u5668\u5b66\u4e60\u201d\u3002\u56fe 3 \u663e\u793a\u4e86\u4f20\u7edf\u673a\u5668\u5b66\u4e60\u7684\u4e00\u4e9b\u65b9\u6cd5\u3002\u6b64\u7c7b\u6a21\u578b\u53ef\u4ee5\u4f7f\u7528\u5404\u79cd\u8f6f\u4ef6\u5305\u6765\u8bad\u7ec3\uff0c\u5305\u62ec Python [18] \u4e2d\u7684 scikit-learn\u3001R [19] \u4e2d\u7684 caret \u548c Julia [20] \u4e2d\u7684 MLJ\u3002<\/span><\/section>\n
 <\/span><\/section>\n

\n
\u88681  \u4e0d\u540c\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u7684\u6bd4\u8f83<\/span><\/strong><\/section>\n
<\/span><\/section>\n
<\/section>\n
<\/section>\n
<\/section>\n
<\/section>\n

<\/section>\n<\/h2>\n
\u5f53\u5f00\u53d1\u7528\u4e8e\u751f\u7269\u6570\u636e\u7684\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\uff0c\u4e3a\u7ed9\u5b9a\u4efb\u52a1\u5bfb\u627e\u6700\u5408\u9002\u65b9\u6cd5\u65f6\uff0c\u901a\u5e38\u5e94\u5c06\u4f20\u7edf\u673a\u5668\u5b66\u4e60\u89c6\u4e3a\u7b2c\u4e00\u4e2a\u63a2\u7d22\u9886\u57df<\/span><\/strong>\u3002\u6df1\u5ea6\u5b66\u4e60\u7684\u786e\u662f\u4e00\u79cd\u5f3a\u5927\u4e14\u76ee\u524d\u65e0\u7591\u6d41\u884c\u7684\u5de5\u5177\u3002\u4f46\u662f\uff0c\u6df1\u5ea6\u5b66\u4e60\u5728\u5176\u64c5\u957f\u7684\u5e94\u7528\u9886\u57df\u4ecd\u7136\u6709\u4e09\u4e2a\u8981\u6c42\uff1a\u6709\u5927\u91cf\u53ef\u7528\u6570\u636e<\/span>\uff08\u4f8b\u5982\uff0c\u6570\u767e\u4e07\u4e2a\u6570\u636e\u70b9\uff09<\/span>\u3001\u6bcf\u4e2a\u6570\u636e\u70b9\u6709\u5f88\u591a\u7279\u5f81\u3001\u7279\u5f81\u9ad8\u5ea6\u7ed3\u6784\u5316<\/span>\uff08\u7279\u5f81\u5f7c\u6b64\u4e4b\u95f4\u6709\u660e\u786e\u7684\u5173\u7cfb\uff0c\u4f8b\u5982\u56fe\u50cf\u4e2d\u7684\u76f8\u90bb\u50cf\u7d20\uff09<\/span>[21]\u3002\u751f\u7269\u6570\u636e\u53ef\u4ee5\u6ee1\u8db3\u8fd9\u4e9b\u8981\u6c42\uff0c\u5e76\u4e14\u6df1\u5ea6\u5b66\u4e60\u5df2\u6210\u529f\u5e94\u7528\u7684\u751f\u7269\u6570\u636e\u7684\u4f8b\u5b50\u5305\u62ec DNA\u3001RNA \u548c\u86cb\u767d\u8d28\u5e8f\u5217 [22, 23] \u4ee5\u53ca\u663e\u5fae\u955c\u56fe\u50cf [24, 25] \u7b49\u6570\u636e\u3002\u7136\u800c\uff0c\u5373\u4f7f\u6ee1\u8db3\u5176\u5b83\u4e24\u4e2a\u8981\u6c42\uff0c\u5bf9\u5927\u91cf\u6570\u636e\u7684\u8981\u6c42\u4e5f\u53ef\u80fd\u4f7f\u6df1\u5ea6\u5b66\u4e60\u6210\u4e3a\u4e00\u4e2a\u7cdf\u7cd5\u7684\u9009\u62e9\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u4e0e\u4f20\u7edf\u65b9\u6cd5\u76f8\u6bd4\uff0c\u6df1\u5ea6\u5b66\u4e60\u5f00\u53d1\u548c\u6d4b\u8bd5\u7ed9\u5b9a\u95ee\u9898\u7684\u901f\u5ea6\u8981\u6162\u5f97\u591a\u3002\u4e0e\u652f\u6301\u5411\u91cf\u673a <\/span>(SVM) <\/span>\u548c\u968f\u673a\u68ee\u6797 [27] \u7b49\u4f20\u7edf\u6a21\u578b\u76f8\u6bd4\uff0c\u5f00\u53d1\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\u7684\u67b6\u6784\u7136\u540e\u5bf9\u5176\u8fdb\u884c\u8bad\u7ec3\u53ef\u80fd\u662f\u4e00\u9879\u8017\u65f6\u4e14\u8ba1\u7b97\u6210\u672c\u9ad8\u7684\u4efb\u52a1 [26]\u3002\u5c3d\u7ba1\u5b58\u5728\u4e00\u4e9b\u65b9\u6cd5\uff0c\u4f46\u5bf9\u4e8e\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\uff0c\u4f30\u8ba1\u7279\u5f81\u91cd\u8981\u6027 [28]\uff08\u5373\u6bcf\u4e2a\u7279\u5f81\u5bf9\u9884\u6d4b\u7684\u91cd\u8981\u6027\uff09\u6216\u6a21\u578b\u9884\u6d4b\u7684\u7f6e\u4fe1\u5ea6\u4ecd\u7136\u4e0d\u662f\u4e00\u4ef6\u5bb9\u6613\u7684\u4e8b<\/strong> [1, 28, 29 ]\uff0c\u8fd9\u4e24\u8005\u5728\u751f\u7269\u73af\u5883\u4e2d\u901a\u5e38\u662f\u5fc5\u4e0d\u53ef\u5c11\u7684\u3002\u5373\u4f7f\u5bf9\u4e8e\u7279\u5b9a\u7684\uff0c\u6df1\u5ea6\u5b66\u4e60\u5728\u6280\u672f\u4e0a\u4f3c\u4e4e\u53ef\u884c\u7684\u751f\u7269\u9884\u6d4b\u4efb\u52a1\uff0c\u8c28\u614e\u7684\u505a\u6cd5\u901a\u5e38\u8fd8\u662f\uff1a\u8bad\u7ec3\u4f20\u7edf\u65b9\u6cd5<\/span>\uff08\u5982\u679c\u53ef\u80fd\u7684\u8bdd\uff09<\/span>\uff0c\u518d\u5c06\u5176\u4e0e\u57fa\u4e8e\u795e\u7ecf\u7f51\u7edc\u7684\u6a21\u578b\u8fdb\u884c\u6bd4\u8f83[30]\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u4f20\u7edf\u65b9\u6cd5\u901a\u5e38\u671f\u671b\u6570\u636e\u96c6\u4e2d\u7684\u6bcf\u4e2a\u793a\u4f8b\u5177\u6709\u76f8\u540c\u6570\u91cf\u7684\u7279\u5f81\uff0c\u56e0\u6b64\u8fd9\u5e76\u4e0d\u603b\u662f\u53ef\u884c\u7684\u3002\u4e00\u4e2a\u660e\u663e\u7684\u751f\u7269\u5b66\u793a\u4f8b\u662f\uff0c\u5728\u4f7f\u7528\u86cb\u767d\u8d28\u3001RNA \u6216 DNA \u5e8f\u5217\u7684\u65f6\u5019\uff0c\u6bcf\u4e2a\u793a\u4f8b\u5177\u6709\u4e0d\u540c\u7684\u957f\u5ea6\u3002\u8981\u5bf9\u8fd9\u4e9b\u6570\u636e\u4f7f\u7528\u4f20\u7edf\u65b9\u6cd5\uff0c\u53ef\u4ee5\u4f7f\u7528\u7b80\u5355\u7684\u6280\u672f<\/span>\uff08\u4f8b\u5982\u586b\u5145\u548c\u52a0\u7a97\uff09<\/span>\u66f4\u6539\u6570\u636e\uff0c\u4f7f\u5b83\u4eec\u7684\u5927\u5c0f\u90fd\u76f8\u540c\u3002\u201c\u586b\u5145\u201d<\/span>\uff08padding\uff09<\/span>\u610f\u5473\u7740\u5bf9\u6bcf\u4e2a\u793a\u4f8b\u6dfb\u52a0\u5305\u542b\u96f6\u7684\u9644\u52a0\u503c\uff0c\u76f4\u5230\u5b83\u4e0e\u6570\u636e\u96c6\u4e2d\u6700\u5927\u7684\u793a\u4f8b\u5927\u5c0f\u76f8\u540c\u3002\u76f8\u6bd4\u4e4b\u4e0b\uff0c\u52a0\u7a97<\/span>\uff08windowing\uff09<\/span>\u5c06\u5355\u4e2a\u793a\u4f8b\u7f29\u77ed\u5230\u7ed9\u5b9a\u7684\u5927\u5c0f<\/span>\uff08\u4f8b\u5982\uff0c\u5728\u4e00\u4e2a\u5e8f\u5217\u957f\u5ea6\u81f3\u5c11\u4e3a100\u7684\u86cb\u767d\u8d28\u5e8f\u5217\u6570\u636e\u96c6\u4e2d\uff0c\u53ea\u4f7f\u7528\u6bcf\u4e2a\u86cb\u767d\u8d28\u7684\u524d 100 \u4e2a\u6b8b\u57fa\uff09<\/span>\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u4f7f\u7528\u5206\u7c7b\u548c\u56de\u5f52\u6a21\u578b<\/span><\/strong><\/span>\u3002\u5bf9\u4e8e\u5982\u56fe3a\u6240\u793a\u7684\u56de\u5f52\u95ee\u9898\uff0c\u5cad\u56de\u5f52<\/span>\uff08\u5177\u6709\u6b63\u5219\u5316\u9879\u7684\u7ebf\u6027\u56de\u5f52\uff09<\/span>\u901a\u5e38\u662f\u5f00\u53d1\u6a21\u578b\u7684\u826f\u597d\u8d77\u70b9\uff0c\u56e0\u4e3a\u5b83\u53ef\u4ee5\u4e3a\u7ed9\u5b9a\u4efb\u52a1\u63d0\u4f9b\u5feb\u901f\u4e14\u6613\u4e8e\u7406\u89e3\u7684\u57fa\u51c6\u3002\u5f53\u5e0c\u671b\u6a21\u578b\u4f9d\u8d56\u53ef\u7528\u6570\u636e\u4e2d\u7684\u6700\u5c11\u7279\u5f81\u65f6\uff0c\u7ebf\u6027\u56de\u5f52\u7684\u5176\u4ed6\u53d8\u4f53<\/span>\uff08\u4f8b\u5982 LASSO \u56de\u5f52 [31] \u548c\u5f39\u6027\u7f51\u7edc\u56de\u5f52 [32]\uff09<\/span>\u4e5f\u503c\u5f97\u8003\u8651\u3002\u4e0d\u5e78\u7684\u662f\uff0c\u6570\u636e\u4e2d\u7279\u5f81\u4e4b\u95f4\u7684\u5173\u7cfb\u901a\u5e38\u662f\u975e\u7ebf\u6027\u7684\u3002\u5bf9\u4e8e\u8fd9\u4e9b\u60c5\u51b5\uff0c\u4f7f\u7528\u8bf8\u5982SVM<\/span>\uff08support vector machine\uff0c\u652f\u6301\u5411\u91cf\u673a\uff09<\/span>\u4e4b\u7c7b\u7684\u6a21\u578b\u901a\u5e38\u662f\u66f4\u5408\u9002\u7684\u9009\u62e9 [33]\u3002SVM \u662f\u4e00\u79cd\u5f3a\u5927\u7684\u56de\u5f52\u548c\u5206\u7c7b\u6a21\u578b\uff0c\u5b83\u4f7f\u7528\u6838\u51fd\u6570<\/span>\uff08kernel function\uff09<\/span>\u5c06\u4e0d\u53ef\u5206\u7684\u95ee\u9898\u8f6c\u6362\u4e3a\u66f4\u5bb9\u6613\u89e3\u51b3\u7684\u53ef\u5206\u95ee\u9898\u3002SVM \u53ef\u7528\u4e8e\u6267\u884c\u7ebf\u6027\u56de\u5f52\u548c\u975e\u7ebf\u6027\u56de\u5f52\uff0c\u5177\u4f53\u53d6\u51b3\u4e8e\u4f7f\u7528\u7684\u6838\u51fd\u6570 [34-37] \u3002\u5f00\u53d1\u6a21\u578b\u7684\u4e00\u4e2a\u597d\u65b9\u6cd5\u662f\u8bad\u7ec3\u4e00\u4e2a\u7ebf\u6027 SVM \u548c\u4e00\u4e2a\u5e26\u6709\u5f84\u5411\u57fa\u51fd\u6570\u6838\u7684 SVM<\/span>\uff08\u4e00\u79cd\u901a\u7528\u975e\u7ebf\u6027\u7c7b\u578b\u7684 SVM\uff09<\/span>\u6765\u91cf\u5316\u53ef\u4ee5\u4ece\u975e\u7ebf\u6027\u51fd\u6570\u4e2d\u83b7\u5f97\u7684\u589e\u76ca<\/span>\uff08\u5982\u679c\u6709\u7684\u8bdd\uff09<\/span>\u3002\u975e\u7ebf\u6027\u65b9\u6cd5\u53ef\u4ee5\u63d0\u4f9b\u66f4\u5f3a\u5927\u7684\u6a21\u578b\uff0c\u4f46\u4ee3\u4ef7\u662f\u4e0d\u5bb9\u6613\u89e3\u91ca\u54ea\u4e9b\u7279\u5f81\u5f71\u54cd\u6a21\u578b\u3002\u8fd9\u5c31\u662f\u524d\u9762\u6240\u63d0\u5230\u7684\u504f\u5dee-\u65b9\u5dee\u6743\u8861\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u56de\u5f52\u4e2d\u5e38\u7528\u7684\u8bb8\u591a\u6a21\u578b\u4e5f\u7528\u4e8e\u5206\u7c7b\u3002\u8bad\u7ec3\u7ebf\u6027 SVM \u548c\u5e26\u6709\u5f84\u5411\u57fa\u51fd\u6570\u6838\u7684 SVM \u4e5f\u662f\u5206\u7c7b\u4efb\u52a1\u7684\u4e00\u4e2a\u5f88\u597d\u7684\u9ed8\u8ba4\u8d77\u70b9\u3002\u53e6\u4e00\u79cd\u53ef\u4ee5\u5c1d\u8bd5\u7684\u65b9\u6cd5\u662fk \u6700\u8fd1\u90bb<\/span>\uff08k nearest neighbours\uff09<\/span>\u5206\u7c7b[38]\u3002\u4f5c\u4e3a\u6700\u7b80\u5355\u7684\u5206\u7c7b\u65b9\u6cd5\u4e4b\u4e00\uff0ck \u6700\u8fd1\u90bb\u5206\u7c7b\u63d0\u4f9b\u4e86\u4e00\u4e2a\u6709\u7528\u7684\u57fa\u7ebf\u6027\u80fd\u6807\u8bb0\uff0c\u53ef\u4ee5\u4e0e\u5176\u4ed6\u66f4\u590d\u6742\u7684\u6a21\u578b<\/span>\uff08\u4f8b\u5982 SVM\uff09<\/span>\u8fdb\u884c\u6bd4\u8f83\u3002\u53e6\u4e00\u7c7b\u7a33\u5065\u7684\u975e\u7ebf\u6027\u65b9\u6cd5\u662f\u57fa\u4e8e\u96c6\u6210<\/span>\uff08ensemble\uff09<\/span>\u7684\u6a21\u578b\uff0c\u4f8b\u5982\u968f\u673a\u68ee\u6797 [39] \u548c XGBoost [40, 41]\u3002\u8fd9\u4e24\u79cd\u65b9\u6cd5\u90fd\u662f\u5f3a\u5927\u7684\u975e\u7ebf\u6027\u6a21\u578b\uff0c\u597d\u5904\u5728\u4e8e\u80fd\u989d\u5916\u63d0\u4f9b\u7279\u5f81\u91cd\u8981\u6027\u7684\u4f30\u8ba1\uff0c\u5e76\u4e14\u901a\u5e38\u9700\u8981\u6700\u5c11\u7684\u8d85\u53c2\u6570\u8c03\u6574\u3002\u5982\u679c\u67d0\u4e2a\u751f\u7269\u5b66\u4efb\u52a1\u9700\u8981\u4e86\u89e3\u54ea\u4e9b\u7279\u5f81\u5bf9\u9884\u6d4b\u7684\u8d21\u732e\u6700\u5927\u624d\u80fd\u7406\u89e3\uff0c\u90a3\u4e48\u8fd9\u4e9b\u6a21\u578b\u5c31\u662f\u4e00\u4e2a\u4e0d\u9519\u7684\u9009\u62e9\uff0c\u6bd5\u7adf\u5b83\u4eec\u80fd\u5206\u914d\u7279\u5f81\u91cd\u8981\u6027\u503c\uff0c\u6709\u51b3\u7b56\u6811\u7ed3\u6784\uff0c<\/span><\/section>\n
       <\/span><\/section>\n
\u5bf9\u4e8e\u5206\u7c7b\u548c\u56de\u5f52\uff0c\u8bb8\u591a\u53ef\u7528\u7684\u6a21\u578b\u5f80\u5f80\u5177\u6709\u4ee4\u4eba\u773c\u82b1\u7f2d\u4e71\u7684\u7279\u70b9\u548c\u53d8\u4f53\u3002\u8bd5\u56fe\u5148\u9a8c\u5730\u9884\u6d4b\u7279\u5b9a\u65b9\u6cd5\u5bf9\u7279\u5b9a\u95ee\u9898\u7684\u9002\u7528\u7a0b\u5ea6\u53ef\u80fd\u9760\u4e0d\u4f4f\uff0c\u800c\u91c7\u7528\u7ecf\u9a8c\u3001\u53cd\u590d\u8bd5\u9a8c\u7684\u65b9\u6cd5\u6765\u5bfb\u627e\u6700\u4f73\u6a21\u578b\u901a\u5e38\u662f\u6700\u8c28\u614e\u7684\u65b9\u6cd5\u3002\u4f7f\u7528\u73b0\u4ee3\u673a\u5668\u5b66\u4e60\u51fd\u6570\u2014\u2014\u4f8b\u5982 scikit-learn [18]\u2014\u2014\u5728\u8fd9\u4e9b\u6a21\u578b\u53d8\u4f53\u4e4b\u95f4\u8fdb\u884c\u66f4\u6539\uff0c\u901a\u5e38\u53ea\u9700\u8981\u66f4\u6539\u4e00\u884c\u4ee3\u7801\u3002\u56e0\u6b64\uff0c\u9009\u62e9\u6700\u4f73\u65b9\u6cd5\u7684\u4e00\u4e2a\u597d\u7684\u603b\u4f53\u7b56\u7565\u662f\uff1a\u8bad\u7ec3\u548c\u4f18\u5316\u4e0a\u8ff0\u5404\u79cd\u65b9\u6cd5\uff0c\u9009\u62e9\u5728\u9a8c\u8bc1\u96c6\u4e0a\u6027\u80fd\u6700\u597d\u7684\u90a3\u4e2a\uff0c\u6700\u7ec8\u6bd4\u8f83\u5b83\u4eec\u5728\u5355\u72ec\u7684\u6d4b\u8bd5\u96c6\u4e0a\u7684\u6027\u80fd\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u4f7f\u7528\u805a\u7c7b\u6a21\u578b\u3002\u805a\u7c7b\u7b97\u6cd5<\/span>\uff08\u56fe 3e\uff09<\/span>\u7684\u4f7f\u7528\u5728\u751f\u7269\u5b66\u4e2d\u5f88\u666e\u904d [42\u300143]\u3002k\u5e73\u5747<\/span>\uff08k-means\uff09<\/span>\u662f\u4e00\u79cd\u5f3a\u5927\u7684\u901a\u7528\u805a\u7c7b\u65b9\u6cd5\uff0c\u4e0e\u8bb8\u591a\u5176\u4ed6\u805a\u7c7b\u7b97\u6cd5\u4e00\u6837\uff0c\u5b83\u9700\u8981\u5c06\u805a\u7c7b\u6570\u8bbe\u7f6e\u4e3a\u8d85\u53c2\u6570 [44]\u3002DBSCAN \u662f\u4e00\u79cd\u66ff\u4ee3\u65b9\u6cd5\uff0c\u5b83\u4e0d\u9700\u8981\u9884\u5b9a\u4e49\u96c6\u7fa4\u7684\u6570\u91cf\uff0c\u4f46\u4ee3\u4ef7\u662f\u9700\u8981\u8bbe\u7f6e\u5176\u4ed6\u8d85\u53c2\u6570 [45] \u3002\u4e5f\u53ef\u4ee5\u5728\u805a\u7c7b\u4e4b\u524d\u6267\u884c\u964d\u7ef4\uff0c\u4ee5\u63d0\u9ad8\u5728\u5177\u6709\u5927\u91cf\u7279\u5f81\u7684\u6570\u636e\u96c6\u4e0a\u7684\u6027\u80fd\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u964d\u7ef4<\/span><\/strong><\/span>\u3002\u964d\u7ef4<\/span>\uff08dimensionality reduction\uff09<\/span>\u6280\u672f\u7528\u4e8e\u5c06\u5177\u6709\u5927\u91cf\u5c5e\u6027<\/span>\uff08\u6216\u7ef4\u5ea6\uff09<\/span>\u7684\u6570\u636e\u8f6c\u6362\u4e3a\u4f4e\u7ef4\u5f62\u5f0f\uff0c\u540c\u65f6\u5c3d\u53ef\u80fd\u4fdd\u7559\u6570\u636e\u70b9\u4e4b\u95f4\u7684\u4e0d\u540c\u5173\u7cfb\u3002\u4f8b\u5982\uff0c\u76f8\u4f3c\u7684\u6570\u636e\u70b9<\/span>\uff08\u4f8b\u5982\uff0c\u4e24\u4e2a\u540c\u6e90\u86cb\u767d\u8d28\u5e8f\u5217\uff09<\/span>\u5728\u5176\u4f4e\u7ef4\u5f62\u5f0f\u4e0a\u4e5f\u5e94\u8be5\u76f8\u4f3c\uff0c\u800c\u4e0d\u540c\u7684\u6570\u636e\u70b9<\/span>\uff08\u4f8b\u5982\uff0c\u4e0d\u76f8\u5173\u7684\u86cb\u767d\u8d28\u5e8f\u5217\uff09<\/span>\u5e94\u8be5\u4fdd\u6301\u4e0d\u76f8\u4f3c[46,47]\u3002\u901a\u5e38\u9009\u62e9\u4e24\u4e2a\u6216\u4e09\u4e2a\u7ef4\u5ea6\u6765\u5141\u8bb8\u5728\u4e00\u7ec4\u8f74\u4e0a\u53ef\u89c6\u5316\u6570\u636e\uff0c\u5c3d\u7ba1\u66f4\u591a\u7684\u7ef4\u5ea6\u4e5f\u7528\u4e8e\u673a\u5668\u5b66\u4e60\u3002\u8fd9\u4e9b\u6280\u672f\u5305\u62ec\u6570\u636e\u7684\u7ebf\u6027\u548c\u975e\u7ebf\u6027\u53d8\u6362\u3002\u751f\u7269\u5b66\u4e2d\u5e38\u7528\u7684\u4f8b\u5b50\u5305\u62ec\u4e3b\u6210\u5206\u5206\u6790<\/span>\uff08principle component analysis\uff0cPCA\uff0c\u5982\u56fe3\uff09<\/span>\u3001UMAP<\/span>\uff08Uniform Manifold Approximation and Projection\uff09<\/span>\u548ct-SNE<\/span>\uff08t-distributed stochastic neighbour embedding\uff09<\/span>\u3002\u751f\u7269\u5b66\u6570\u636e\u5904\u7406\u4e2d\u5177\u4f53\u4f7f\u7528\u7684\u6280\u672f\u53d6\u51b3\u4e8e\u60c5\u51b5\uff1aPCA \u4fdd\u7559\u6570\u636e\u70b9\u4e4b\u95f4\u7684\u5168\u5c40\u5173\u7cfb\u5e76\u4e14\u662f\u53ef\u89e3\u91ca\u7684\uff0c\u56e0\u4e3a\u6bcf\u4e2a\u7ec4\u4ef6\u90fd\u662f\u4e00\u4e2a\u8f93\u5165\u7279\u5f81\u7684\u7ebf\u6027\u7ec4\u5408\uff0c\u8fd9\u610f\u5473\u7740\u5f88\u5bb9\u6613\u7406\u89e3\u54ea\u4e9b\u7279\u5f81\u4f1a\u5bfc\u81f4\u6570\u636e\u7684\u591a\u6837\u6027\u3002t-SNE \u66f4\u597d\u5730\u4fdd\u7559\u4e86\u6570\u636e\u70b9\u4e4b\u95f4\u7684\u5c40\u90e8\u5173\u7cfb\uff0c\u662f\u4e00\u79cd\u7075\u6d3b\u7684\u65b9\u6cd5\uff0c\u53ef\u4ee5\u63ed\u793a\u590d\u6742\u6570\u636e\u96c6\u4e2d\u7684\u7ed3\u6784\u3002\u5e94\u7528\u5305\u62ec\u7528\u4e8e t-SNE[49] \u7684\u5355\u7ec6\u80de\u8f6c\u5f55\u7ec4\u5b66\u548c\u7528\u4e8e\u4e3b\u6210\u5206\u5206\u6790\u7684\u5206\u5b50\u52a8\u529b\u5b66\u8f68\u8ff9\u5206\u6790\u3002<\/span><\/section>\n
 <\/span><\/section>\n
<\/section>\n

\u56fe3. \u4f20\u7edf\u7684\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u3002(a) <\/strong>\u56de\u5f52\u5bfb\u627e\u56e0\u53d8\u91cf\uff08\u89c2\u5bdf\u5230\u7684\u5c5e\u6027\uff09\u548c\u4e00\u4e2a\u6216\u591a\u4e2a\u81ea\u53d8\u91cf\uff08\u7279\u5f81\uff09\u4e4b\u95f4\u7684\u5173\u7cfb\uff1b\u4f8b\u5982\uff0c\u4ece\u4e00\u4e2a\u4eba\u7684\u8eab\u9ad8\u9884\u6d4b\u4e00\u4e2a\u4eba\u7684\u4f53\u91cd\u3002(b) <\/strong>\u652f\u6301\u5411\u91cf\u673a (SVM) \u8f6c\u6362\u539f\u59cb\u8f93\u5165\u6570\u636e\uff0c\u4ee5\u4fbf\u5728\u5b83\u4eec\u7684\u8f6c\u6362\u7248\u672c\uff08\u79f0\u4e3a\u201c\u6f5c\u5728\u8868\u793a\u201d\uff09\u4e2d\uff0c\u5c5e\u4e8e\u4e0d\u540c\u7c7b\u522b\u7684\u6570\u636e\u88ab\u5c3d\u53ef\u80fd\u5bbd\u7684\u660e\u663e\u95f4\u9699\u5212\u5206\u3002\u8fd9\u91cc\u5c55\u793a\u4e86\u5bf9\u86cb\u767d\u8d28\u662f\u6709\u5e8f\u8fd8\u662f\u65e0\u5e8f\u7684\u9884\u6d4b\uff0c\u6a2a\u8f74\u4e0e\u7eb5\u8f74\u4ee3\u8868\u8f6c\u6362\u6570\u636e\u7684\u7ef4\u5ea6\u3002(c) <\/strong>\u68af\u5ea6\u63d0\u5347\u4f7f\u7528\u4e00\u7ec4\u5f31\u9884\u6d4b\u6a21\u578b\uff08\u901a\u5e38\u662f\u51b3\u7b56\u6811\uff09\u6765\u8fdb\u884c\u9884\u6d4b\u3002\u4f8b\u5982\uff0c\u6d3b\u6027\u836f\u7269\u53ef\u4ee5\u4ece\u5206\u5b50\u63cf\u8ff0\u7b26\uff08\u4f8b\u5982\u5206\u5b50\u91cf\u548c\u7279\u5b9a\u5316\u5b66\u57fa\u56e2\u7684\u5b58\u5728\uff09\u4e2d\u9884\u6d4b\u3002\u5404\u4e2a\u9884\u6d4b\u5668\u4ee5\u9636\u6bb5\u65b9\u5f0f\u7ec4\u5408\u4ee5\u8fdb\u884c\u6700\u7ec8\u9884\u6d4b\u3002(d) <\/strong>\u4e3b\u6210\u5206\u5206\u6790 (PCA) \u4f1a\u53d1\u73b0\u4e00\u7cfb\u5217\u7279\u5f81\u7ec4\u5408\uff0c\u8fd9\u4e9b\u7279\u5f81\u7ec4\u5408\u5728\u5f7c\u6b64\u6b63\u4ea4\u7684\u60c5\u51b5\u4e0b\u6700\u80fd\u63cf\u8ff0\u6570\u636e\u3002\u5b83\u901a\u5e38\u7528\u4e8e\u964d\u7ef4\u3002\u5728\u201c\u8eab\u9ad8\u4e0e\u4f53\u91cd\u201d\u7684\u4f8b\u5b50\u4e2d\uff0c\u5bf9\u5e94\u4e8e\u8eab\u9ad8\u548c\u4f53\u91cd\u7684\u7ebf\u6027\u7ec4\u5408\u7684\u7b2c\u4e00\u4e3b\u6210\u5206 (first principal component\uff0cPC1) \u63cf\u8ff0\u4e86\u5f3a\u6b63\u76f8\u5173\uff0c\u800c PC2 \u53ef\u80fd\u63cf\u8ff0\u4e86\u4e0e\u8eab\u9ad8\u4e0e\u4f53\u91cd\u4e0d\u5f3a\u76f8\u5173\u7684\u5176\u4ed6\u53d8\u91cf\uff0c\u4f8b\u5982\u4f53\u8102\u767e\u5206\u6bd4\u6216\u808c\u8089\u8d28\u91cf\u767e\u5206\u6bd4\u3002(e)<\/strong> \u805a\u7c7b\u4f7f\u7528\u5404\u79cd\u7b97\u6cd5\u4e2d\u7684\u4e00\u79cd\uff0c\u6765\u5bf9\u76f8\u4f3c\u7684\u5bf9\u8c61\u8fdb\u884c\u5206\u7ec4\uff08\u4f8b\u5982\uff0c\u6839\u636e\u57fa\u56e0\u8868\u8fbe\u8c31\u5bf9\u7ec6\u80de\u7c7b\u578b\u8fdb\u884c\u5206\u7ec4\uff09\u3002<\/span><\/p>\n

 <\/span><\/section>\n


<\/p>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

\u4eba\u5de5\u795e\u7ecf\u7f51\u7edc<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n


<\/p>\n

\u4eba\u5de5\u795e\u7ecf\u7f51\u7edc<\/span>\uff08artifical neural networks\uff09<\/span>\u6a21\u578b\u4f1a\u6709\u8fd9\u4e2a\u540d\u5b57\uff0c\u662f\u56e0\u4e3a\u8981\u62df\u5408\u7684\u6570\u5b66\u6a21\u578b\u7684\u5f62\u5f0f\u53d7\u5230\u5927\u8111\u4e2d\u795e\u7ecf\u5143\u7684\u8fde\u901a\u6027\u548c\u884c\u4e3a\u7684\u542f\u53d1\uff0c\u4ee5\u53ca\u8fd9\u4e2a\u6a21\u578b\u6700\u521d\u65e8\u5728\u4e86\u89e3\u5927\u8111\u529f\u80fd [50]\u3002\u7136\u800c\uff0c\u6570\u636e\u79d1\u5b66\u4e2d\u5e38\u7528\u7684\u795e\u7ecf\u7f51\u7edc\u4f5c\u4e3a\u5927\u8111\u6a21\u578b\u5df2\u7ecf\u8fc7\u65f6\uff0c\u73b0\u5728\uff0c\u5b83\u53ea\u662f\u673a\u5668\u5b66\u4e60\u6a21\u578b\uff0c\u53ef\u4ee5\u5728\u67d0\u4e9b\u5e94\u7528\u4e2d\u63d0\u4f9b\u6700\u5148\u8fdb\u6027\u80fd\u3002\u7531\u4e8e\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\u7684\u67b6\u6784\u548c\u8bad\u7ec3\u7684\u5feb\u901f\u53d1\u5c55\uff0c\u8fd1\u51e0\u5341\u5e74\u6765\uff0c\u4eba\u4eec\u5bf9\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u7684\u5174\u8da3\u4e0d\u65ad\u589e\u957f [26]\u3002\u672c\u8282\u5c06\u63cf\u8ff0\u57fa\u672c\u7684\u795e\u7ecf\u7f51\u7edc\uff0c\u4ee5\u53ca\u5e7f\u6cdb\u7528\u4e8e\u751f\u7269\u5b66\u7814\u7a76\u7684\u53d8\u4f53\u3002\u5176\u4e2d\u4e00\u4e9b\u5982\u56fe4\u6240\u793a\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u795e\u7ecf\u7f51\u7edc\u7684\u57fa\u672c\u539f\u7406<\/span><\/strong><\/span>\u3002\u795e\u7ecf\u7f51\u7edc\u7684\u4e00\u4e2a\u5173\u952e\u7279\u6027\u662f\u5b83\u4eec\u662f\u901a\u7528\u51fd\u6570\u903c\u8fd1\u5668\uff0c\u8fd9\u610f\u5473\u7740\uff0c\u53ea\u9700\u5f88\u5c11\u7684\u5047\u8bbe\uff0c\u6b63\u786e\u914d\u7f6e\u7684\u795e\u7ecf\u7f51\u7edc\u5c31\u53ef\u4ee5\u5c06\u4efb\u4f55\u6570\u5b66\u51fd\u6570\u903c\u8fd1\u4efb\u610f\u7cbe\u5ea6\u6c34\u5e73<\/strong>\u3002\u6362\u53e5\u8bdd\u8bf4\uff0c\u5982\u679c\u4efb\u4f55\u8fc7\u7a0b<\/span>\uff08\u751f\u7269\u7684\u6216\u5176\u4ed6\u7684\uff09<\/span>\u53ef\u4ee5\u88ab\u8ba4\u4e3a\u662f\u4e00\u7ec4\u53d8\u91cf\u7684\u67d0\u79cd\u51fd\u6570\uff0c\u90a3\u4e48\u8be5\u8fc7\u7a0b\u53ef\u4ee5\u88ab\u5efa\u6a21\u5230\u4efb\u610f\u7a0b\u5ea6\u7684\u51c6\u786e\u5ea6\uff0c\u4ec5\u53d7\u6a21\u578b\u7684\u5927\u5c0f\u6216\u590d\u6742\u6027\u7684\u63a7\u5236\u3002\u4e0a\u8ff0\u5bf9\u901a\u7528\u903c\u8fd1\u7684\u5b9a\u4e49\u5728\u6570\u5b66\u4e0a\u5e76\u4e0d\u4e25\u8c28\uff0c\u4f46\u786e\u5b9e\u7a81\u51fa\u4e86\u5bf9\u795e\u7ecf\u7f51\u7edc\u7684\u5174\u8da3\u6301\u7eed\u6570\u5341\u5e74\u7684\u4e00\u4e2a\u539f\u56e0\u3002\u7136\u800c\uff0c\u8fd9\u79cd\u4fdd\u8bc1\u5e76\u6ca1\u6709\u63d0\u4f9b\u4e00\u79cd\u627e\u5230\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u7684\u6700\u4f73\u53c2\u6570\uff0c\u4e3a\u7ed9\u5b9a\u7684\u6570\u636e\u96c6\u4ea7\u751f\u6700\u4f73\u8fd1\u4f3c\u503c\u7684\u65b9\u6cd5\uff0c\u4e5f\u4e0d\u80fd\u4fdd\u8bc1\u8be5\u6a21\u578b\u5c06\u4e3a\u65b0\u6570\u636e\u63d0\u4f9b\u51c6\u786e\u7684\u9884\u6d4b [51]\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u4eba\u5de5\u795e\u7ecf\u5143\u662f\u6240\u6709\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u7684\u57fa\u77f3\u3002\u4e00\u4e2a\u4eba\u5de5\u795e\u7ecf\u5143\u53ea\u662f\u4e00\u4e2a\u6570\u5b66\u51fd\u6570\uff0c\u4ee5\u7279\u5b9a\u65b9\u5f0f\u5c06\u8f93\u5165\u6620\u5c04<\/span>\uff08\u8f6c\u6362\uff09<\/span>\u4e3a\u8f93\u51fa\u3002\u5355\u4e2a\u4eba\u5de5\u795e\u7ecf\u5143\u63a5\u53d7\u4efb\u610f\u6570\u91cf\u7684\u8f93\u5165\u503c\uff0c\u5bf9\u5b83\u4eec\u5e94\u7528\u7279\u5b9a\u7684\u6570\u5b66\u51fd\u6570\u5e76\u8fd4\u56de\u8f93\u51fa\u503c\u3002\u4f7f\u7528\u7684\u51fd\u6570\u901a\u5e38\u8868\u793a\u4e3a\uff1a<\/span><\/section>\n
 <\/span><\/section>\n
<\/section>\n
 <\/span><\/section>\n
\u5176\u4e2dxi<\/sub><\/span> \u8868\u793a\u5355\u4e2a\u8f93\u5165\u53d8\u91cf\u6216\u7279\u5f81<\/span>\uff08\u6709 n \u4e2a\u8fd9\u6837\u7684\u8f93\u5165\uff09<\/span>\uff0cwi<\/sub><\/span> \u8868\u793a\u8be5\u8f93\u5165\u7684\u53ef\u5b66\u4e60\u6743\u91cd<\/span>\uff08weight\uff09<\/span>\uff0cb \u8868\u793a\u53ef\u5b66\u4e60\u7684\u504f\u5dee\u9879\uff0c\u03c3 \u8868\u793a\u63a5\u53d7\u5355\u4e2a\u8f93\u5165\u5e76\u8fd4\u56de\u5355\u4e2a\u8f93\u5165\u7684\u975e\u7ebf\u6027\u6fc0\u6d3b\u51fd\u6570\u8f93\u51fa\u3002\u521b\u5efa\u7f51\u7edc\u7684\u8fc7\u7a0b\u4e2d\uff0c\u4eba\u5de5\u795e\u7ecf\u5143\u6309\u5c42\u6392\u5217\uff0c\u4e00\u5c42\u7684\u8f93\u51fa\u4f5c\u4e3a\u4e0b\u4e00\u5c42\u7684\u8f93\u5165\u3002\u7f51\u7edc\u7684\u8282\u70b9\u53ef\u4ee5\u88ab\u8ba4\u4e3a\u662f\u6301\u6709\u4e0a\u8ff0\u65b9\u7a0b\u4e2d\u7684 y \u503c\uff0c\u8fd9\u4e9b\u503c\u6210\u4e3a\u4e0b\u4e00\u5c42\u7684 x \u503c\u3002\u4ee5\u4e0b\u5c0f\u8282\u63cf\u8ff0\u4e86\u6392\u5217\u4eba\u5de5\u795e\u7ecf\u5143\u7684\u5404\u79cd\u65b9\u6cd5\uff0c\u8fd9\u4e9b\u5c0f\u8282\u79f0\u4e3a\u201c\u795e\u7ecf\u7f51\u7edc\u67b6\u6784\u201d\u3002\u7ed3\u5408\u4e0d\u540c\u7684\u67b6\u6784\u7c7b\u578b\u4e5f\u5f88\u5e38\u89c1\uff1b\u4f8b\u5982\uff0c\u5728\u7528\u4e8e\u5206\u7c7b\u7684\u5377\u79ef\u795e\u7ecf\u7f51\u7edc<\/span> (CNN) <\/span>\u4e2d\uff0c\u901a\u5e38\u4f7f\u7528\u5168\u8fde\u63a5\u5c42\u6765\u4ea7\u751f\u6700\u7ec8\u7684\u5206\u7c7b\u8f93\u51fa\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u591a\u5c42\u611f\u77e5\u673a<\/span><\/strong><\/span>\uff08multilayer perceptron\uff09<\/span>\u3002\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u6700\u57fa\u672c\u7684\u5e03\u5c40\u662f\u4ee5\u5168\u8fde\u63a5\u65b9\u5f0f\u6392\u5217\u7684\u4eba\u5de5\u795e\u7ecf\u5143\u5c42\uff0c\u5982\u56fe4a\u6240\u793a\u3002\u5728\u8fd9\u4e2a\u5e03\u5c40\u4e2d\uff0c\u56fa\u5b9a\u6570\u91cf\u7684\u201c\u8f93\u5165\u795e\u7ecf\u5143\u201d\u4ee3\u8868\u6839\u636e\u8f93\u5165\u7f51\u7edc\u7684\u6570\u636e\u8ba1\u7b97\u7684\u8f93\u5165\u7279\u5f81\u503c\uff0c\u4e00\u5bf9\u795e\u7ecf\u5143\u4e4b\u95f4\u7684\u6bcf\u4e2a\u8fde\u63a5\u4ee3\u8868\u4e00\u4e2a\u53ef\u8bad\u7ec3\u7684\u6743\u91cd\u53c2\u6570\u3002\u8fd9\u4e9b\u6743\u91cd\u662f\u795e\u7ecf\u7f51\u7edc\u4e2d\u4e3b\u8981\u7684\u53ef\u8c03\u53c2\u6570\uff0c\u4f18\u5316\u8fd9\u4e9b\u6743\u91cd\u5c31\u662f\u795e\u7ecf\u7f51\u7edc\u8bad\u7ec3\u7684\u610f\u4e49\u6240\u5728\u3002\u5728\u7f51\u7edc\u7684\u53e6\u4e00\u7aef\uff0c\u8bb8\u591a\u8f93\u51fa\u795e\u7ecf\u5143\u4ee3\u8868\u7f51\u7edc\u7684\u6700\u7ec8\u8f93\u51fa\u503c\u3002\u8fd9\u79cd\u7f51\u7edc\u5728\u6b63\u786e\u914d\u7f6e\u540e\u53ef\u7528\u4e8e\u5bf9\u8f93\u5165\u505a\u51fa\u590d\u6742\u7684\u5206\u5c42\u51b3\u7b56\uff0c\u56e0\u4e3a\u7ed9\u5b9a\u5c42\u4e2d\u7684\u6bcf\u4e2a\u795e\u7ecf\u5143\u90fd\u63a5\u6536\u6765\u81ea\u524d\u4e00\u5c42\u4e2d\u6240\u6709\u795e\u7ecf\u5143\u7684\u8f93\u5165\u3002\u8fd9\u79cd\u7b80\u5355\u6392\u5217\u7684\u795e\u7ecf\u5143\u5c42\u901a\u5e38\u88ab\u79f0\u4e3a\u201c\u591a\u5c42\u611f\u77e5\u673a\u201d\uff0c\u662f\u7b2c\u4e00\u4e2a\u53ef\u7528\u4e8e\u751f\u7269\u4fe1\u606f\u5b66\u5e94\u7528\u7684\u7f51\u7edc [52, 53]\u3002\u7531\u4e8e\u5176\u8bad\u7ec3\u7684\u7b80\u4fbf\u6027\u548c\u901f\u5ea6\uff0c\u5b83\u4eec\u4eca\u5929\u4ecd\u5e7f\u6cdb\u7528\u4e8e\u8bb8\u591a\u751f\u7269\u5efa\u6a21\u5e94\u7528\u7a0b\u5e8f [13, 54]\u3002\u7136\u800c\uff0c\u5728\u8bb8\u591a\u5176\u4ed6\u5e94\u7528\u4e2d\uff0c\u8fd9\u4e9b\u7b80\u5355\u7684\u67b6\u6784\u5df2\u7ecf\u88ab\u4e0b\u9762\u8ba8\u8bba\u7684\u8f83\u65b0\u7684\u6a21\u578b\u67b6\u6784\u6240\u8d85\u8d8a\uff0c\u5c3d\u7ba1\u5176\u4e2d\u4e00\u4e9b\u8f83\u65b0\u7684\u67b6\u6784\u4ecd\u7136\u7ecf\u5e38\u4f7f\u7528\u5168\u8fde\u63a5\u5c42\u4f5c\u4e3a\u5b50\u7ec4\u4ef6\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u5377\u79ef\u795e\u7ecf\u7f51\u7edc<\/span><\/strong><\/span>\uff08CNN\uff09<\/span>\u3002CNN \u975e\u5e38\u9002\u5408\u56fe\u50cf\u7c7b\u6570\u636e\uff0c\u5f53\u6570\u636e\u5177\u6709\u67d0\u79cd\u7c7b\u578b\u7684\u5c40\u90e8\u7ed3\u6784\uff0c\u5e76\u4e14\u8bc6\u522b\u8fd9\u79cd\u7ed3\u6784\u662f\u5206\u6790\u7684\u5173\u952e\u76ee\u6807\u7684\u65f6\u5019\u3002\u4ee5\u56fe\u50cf\u4e3a\u4f8b\uff0c\u8fd9\u79cd\u5c40\u90e8\u7ed3\u6784\u53ef\u80fd\u4e0e\u89c6\u91ce\u4e2d\u7279\u5b9a\u7c7b\u578b\u7684\u5bf9\u8c61<\/span>\uff08\u4f8b\u5982\uff0c\u663e\u5fae\u955c\u56fe\u50cf\u4e2d\u7684\u7ec6\u80de\uff09<\/span>\u76f8\u5173\uff0c\u5728\u8f93\u5165\u56fe\u50cf\u4e2d\u7531\u7279\u5b9a\u5c40\u90e8\u989c\u8272\u6a21\u5f0f\u548c\/\u6216\u7a7a\u95f4\u63a5\u8fd1\u50cf\u7d20\u4e2d\u7684\u8fb9\u7f18\u8868\u793a\u3002<\/span><\/section>\n
       <\/span><\/section>\n
CNN \u7531\u4e00\u4e2a\u6216\u591a\u4e2a\u5377\u79ef\u5c42\u7ec4\u6210<\/span>\uff08\u89c1\u56fe 4b\uff09<\/span>\uff0c\u5176\u4e2d\u8f93\u51fa\u662f\u5c06\u4e00\u4e2a\u5c0f\u7684\u3001\u4e00\u5c42\u5168\u8fde\u63a5\u7684\u795e\u7ecf\u7f51\u7edc<\/span>\uff08\u79f0\u4e3a\u201c\u8fc7\u6ee4\u5668\u201d\uff08filter\uff09\u6216\u201c\u6838\u201d\uff08kernel\uff09\uff09<\/span>\u5e94\u7528\u4e8e\u8f93\u5165\u4e2d\u7684\u5c40\u90e8\u7279\u5f81\u7684\u7ed3\u679c\u3002\u5982\u679c\u8f93\u5165\u662f\u7c7b\u4f3c\u56fe\u50cf\u7684\uff0c\u8be5\u5c40\u90e8\u533a\u57df\u5c31\u662f\u56fe\u50cf\u4e2d\u7684\u4e00\u5c0f\u5757\u50cf\u7d20\u3002\u5377\u79ef\u5c42\u7684\u8f93\u51fa\u4e5f\u662f\u7c7b\u4f3c\u56fe\u50cf\u7684\u6570\u7ec4\uff0c\u5305\u542b\u6ee4\u6ce2\u5668\u5728\u6574\u4e2a\u8f93\u5165\u4e0a\u201c\u6ed1\u52a8\u201d\u5e76\u5728\u6bcf\u4e2a\u4f4d\u7f6e\u8ba1\u7b97\u8f93\u51fa\u7684\u7ed3\u679c\u3002\u81f3\u5173\u91cd\u8981\u7684\u662f\uff0c\u6240\u6709\u50cf\u7d20\u90fd\u4f7f\u7528\u76f8\u540c\u7684\u8fc7\u6ee4\u5668\uff0c\u5141\u8bb8\u8fc7\u6ee4\u5668\u5b66\u4e60\u8f93\u5165\u6570\u636e\u4e2d\u7684\u5c40\u90e8\u7ed3\u6784\u3002\u5728\u66f4\u6df1\u7684 CNN \u4e2d\uff0c\u4f7f\u7528\u8df3\u8dc3\u8fde\u63a5<\/span>\uff08skip connection\uff09<\/span>\u662f\u5f88\u5e38\u89c1\u7684\uff0c\u9664\u4e86\u901a\u8fc7\u5c42\u4e2d\u7684\u5904\u7406\u5355\u5143\u4e4b\u5916\uff0c\u8fd8\u5141\u8bb8\u8f93\u5165\u4fe1\u53f7\u7ed5\u8fc7\u4e00\u4e2a\u6216\u591a\u4e2a\u5c42\u3002\u8fd9\u79cd\u7c7b\u578b\u7684\u7f51\u7edc\u79f0\u4e3a\u201c\u6b8b\u5dee\u7f51\u7edc\u201d<\/span>\uff08residual network\uff09<\/span>\uff0c\u53ef\u4ee5\u8ba9\u8bad\u7ec3\u66f4\u5feb\u5730\u6536\u655b\u5230\u51c6\u786e\u7684\u89e3\u51b3\u65b9\u6848\u4e0a\u3002<\/span><\/section>\n
       <\/span><\/section>\n
CNN \u7ecf\u8fc7\u914d\u7f6e\uff0c\u5bf9\u4e0d\u540c\u7a7a\u95f4\u7ed3\u6784\u7684\u6570\u636e\u90fd\u53ef\u4ee5\u8fdb\u884c\u6709\u6548\u64cd\u4f5c\u3002\u4f8b\u5982\uff0c\u4e00\u7ef4 CNN \u7684\u8fc7\u6ee4\u5668\u53ea\u4f1a\u5728\u4e00\u4e2a\u65b9\u5411\u4e0a\u6ed1\u52a8<\/span>\uff08\u6bd4\u5982\u4ece\u5de6\u5230\u53f3\uff09<\/span>\uff1b\u8fd9\u79cd\u7c7b\u578b\u7684 CNN \u9002\u7528\u4e8e\u53ea\u6709\u4e00\u4e2a\u7a7a\u95f4\u7ef4\u5ea6\u7684\u6570\u636e<\/span>\uff08\u4f8b\u5982\u6587\u672c\u6216\u751f\u7269\u5e8f\u5217\uff09<\/span>\u3002\u4e8c\u7ef4 CNN \u5bf9\u5177\u6709\u4e24\u4e2a\u7a7a\u95f4\u7ef4\u5ea6\u7684\u6570\u636e\u8fdb\u884c\u64cd\u4f5c\uff0c\u4f8b\u5982\u6570\u5b57\u56fe\u50cf\u3002\u4e09\u7ef4 CNN \u5bf9\u4f53\u79ef\u6570\u636e\u8fdb\u884c\u64cd\u4f5c\uff0c\u4f8b\u5982\u78c1\u5171\u632f\u6210\u50cf\u626b\u63cf\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u5728\u751f\u7269\u5b66\u4e2d\uff0cCNN \u9488\u5bf9\u5404\u79cd\u6570\u636e\u7c7b\u578b\u90fd\u5df2\u53d6\u5f97\u4e86\u91cd\u5927\u6210\u529f\u3002\u86cb\u767d\u8d28\u7ed3\u6784\u9884\u6d4b\u7684\u6700\u65b0\u8fdb\u5c55\u662f\u4f7f\u7528\u76f8\u5173\u86cb\u767d\u8d28\u5e8f\u5217\u4e2d\u6b8b\u57fa\u5bf9\u5171\u540c\u8fdb\u5316\u7684\u4fe1\u606f\uff0c\u6765\u63d0\u53d6\u6b8b\u57fa\u5bf9\u63a5\u89e6\u548c\u8ddd\u79bb\u7684\u4fe1\u606f\uff0c\u4ece\u800c\u80fd\u591f\u4ee5\u524d\u6240\u672a\u6709\u7684\u51c6\u786e\u5ea6\u5efa\u7acb\u5bf9\u4e09\u7ef4\u86cb\u767d\u8d28\u7ed3\u6784\u7684\u9884\u6d4b [23,55]\u3002\u5728\u8fd9\u79cd\u60c5\u51b5\u4e0b\uff0c\u7f51\u7edc\u5b66\u4e60\u6311\u9009\u51fa\u76f4\u63a5\u8026\u5408\u7684\u76f8\u4e92\u4f5c\u7528\uff0c\u5373\u4f7f\u4e00\u4e2a\u5e8f\u5217\u53ea\u6709\u5f88\u5c11\u6216\u6ca1\u6709\u76f8\u5173\u5e8f\u5217\uff0c\u4e5f\u53ef\u4ee5\u5bf9\u5b83\u505a\u51fa\u51c6\u786e\u7684\u9884\u6d4b [56]\u3002CNN \u4e5f\u5df2\u6210\u529f\u5e94\u7528\u4e8e\u8bc6\u522b\u57fa\u56e0\u5e8f\u5217\u6570\u636e\u4e2d\u7684\u53d8\u5f02 [57] \u3001\u4e09\u7ef4\u57fa\u56e0\u7ec4\u6298\u53e0 [58] \u3001DNA-\u86cb\u767d\u8d28\u76f8\u4e92\u4f5c\u7528 [22, 59] \u3001\u4f4e\u6e29\u7535\u5b50\u663e\u5fae\u955c\u56fe\u50cf\u5206\u6790 [60, 61] \u548c\u533b\u5b66\u91cd\u8981\u9886\u57df\u7684\u56fe\u50cf\u5206\u7c7b\u3002\u76ee\u524d\uff0c\u5b83\u4eec\u5728\u4f8b\u5982\u68c0\u6d4b\u6076\u6027\u80bf\u7624\u4e4b\u7c7b\u7684\u9886\u57df\uff0c\u5df2\u7ecf\u53ef\u4ee5\u4e0e\u4eba\u7c7b\u4e13\u5bb6\u7684\u8868\u73b0\u76f8\u5ab2\u7f8e[24,62]\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u5faa\u73af\u795e\u7ecf\u7f51\u7edc<\/span><\/strong><\/span>\uff08RNN\uff09<\/span>\u3002RNN \u6700\u9002\u5408\u4e8e\u6709\u5e8f\u5e8f\u5217<\/span>\uff08ordered sequences\uff09<\/span>\u5f62\u5f0f\u7684\u6570\u636e\uff0c\u5728\u8fd9\u4e9b\u5e8f\u5217\u4e2d\uff0c\u4e00\u4e2a\u70b9\u548c\u4e0b\u4e00\u4e2a\u70b9\u4e4b\u95f4<\/span>\uff08\u81f3\u5c11\u5728\u7406\u8bba\u4e0a\uff09<\/span>\u5b58\u5728\u4e00\u4e9b\u76f8\u5173\u6027\u6216\u76f8\u5173\u6027\u3002\u5b83\u4eec\u5728\u751f\u7269\u5b66\u4e4b\u5916\u7684\u5e94\u7528\uff0c\u53ef\u80fd\u4e3b\u8981\u662f\u5728\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e2d\uff0c\u5176\u4e2d\u6587\u672c\u88ab\u89c6\u4e3a\u4e00\u7cfb\u5217\u5355\u8bcd\u6216\u5b57\u7b26\u3002\u5982\u56fe 4c \u6240\u793a\uff0cRNN \u53ef\u4ee5\u88ab\u8ba4\u4e3a\u662f\u4e00\u4e2a\u795e\u7ecf\u7f51\u7edc\u5c42\u5757\uff0c\u5b83\u5c06\u5e8f\u5217\u4e2d\u6bcf\u4e2a\u6761\u76ee<\/span>\uff08\u6216\u65f6\u95f4\u6b65\u957f\uff09<\/span>\u5bf9\u5e94\u7684\u6570\u636e\u4f5c\u4e3a\u8f93\u5165\uff0c\u5e76\u4e3a\u6bcf\u4e2a\u6761\u76ee\u751f\u6210\u4e00\u4e2a\u8f93\u51fa\uff0c\u8be5\u8f93\u51fa\u4f9d\u8d56\u4e8e\u5148\u524d\u5904\u7406\u8fc7\u7684\u6761\u76ee\u3002\u5b83\u4eec\u8fd8\u53ef\u7528\u4e8e\u751f\u6210\u6574\u4e2a\u5e8f\u5217\u7684\u8868\u793a\uff0c\u8be5\u8868\u793a\u4f20\u9012\u5230\u7f51\u7edc\u7684\u540e\u7eed\u5c42\u4ee5\u751f\u6210\u8f93\u51fa\u3002\u8fd9\u5728\u79d1\u7814\u4e2d\u4f18\u52bf\u5de8\u5927\uff1a\u56e0\u4e3a\u4efb\u4f55\u957f\u5ea6\u7684\u5e8f\u5217\u90fd\u53ef\u4ee5\u8f6c\u6362\u4e3a\u56fa\u5b9a\u5927\u5c0f\u7684\u8868\u793a\u5e76\u8f93\u5165\u5230\u591a\u5c42\u611f\u77e5\u673a<\/strong>\u3002\u5728\u751f\u7269\u5b66\u4e2d\u4f7f\u7528 RNN \u6548\u679c\u6700\u660e\u663e\u7684\u4f8b\u5b50\u662f\u5206\u6790\u57fa\u56e0\u6216\u86cb\u767d\u8d28\u5e8f\u5217\uff1a\u5176\u4efb\u52a1\u5305\u62ec\u4ece\u57fa\u56e0\u5e8f\u5217\u4e2d\u8bc6\u522b\u542f\u52a8\u5b50\u533a\u57df\u3001\u9884\u6d4b\u86cb\u767d\u8d28\u4e8c\u7ea7\u7ed3\u6784\uff0c\u4ee5\u53ca\u5efa\u6a21\u57fa\u56e0\u8868\u8fbe\u6c34\u5e73\u968f\u65f6\u95f4\u7684\u53d8\u5316\uff1b\u5728\u6700\u540e\u4e00\u79cd\u60c5\u51b5\u4e0b\uff0c\u7ed9\u5b9a\u65f6\u95f4\u70b9\u7684\u503c\u5c06\u8ba1\u4e3a\u5e8f\u5217\u4e2d\u7684\u4e00\u4e2a\u6761\u76ee\u3002RNN \u7684\u66f4\u5148\u8fdb\u7684\u53d8\u4f53\uff0c\u957f\u77ed\u671f\u8bb0\u5fc6<\/span>\uff08long short-term memory\uff09<\/span>\u6216\u95e8\u63a7\u5faa\u73af\u5355\u5143<\/span>\uff08gated recurrent unit\uff09<\/span>\uff0c\u5728\u751f\u7269\u5b66\u4e2d\u6709\u8bb8\u591a\u7528\u9014\uff0c\u5305\u62ec\u86cb\u767d\u8d28\u7ed3\u6784\u9884\u6d4b [63, 64] \u3001\u80bd\u8bbe\u8ba1 [65] \u548c\u6839\u636e\u5065\u5eb7\u8bb0\u5f55\u9884\u6d4b\u4e34\u5e8a\u8bca\u65ad [66] \u3002\u8fd9\u4e9b\u66f4\u9ad8\u7ea7\u7684\u65b9\u6cd5\u901a\u5e38\u4e0e CNN \u7ed3\u5408\u4f7f\u7528\uff0c\u53ef\u4ee5\u63d0\u9ad8\u51c6\u786e\u6027 [67]\u3002RNN \u5728\u5206\u6790\u57fa\u4e8e\u5e8f\u5217\u7684\u6570\u636e\u65f6\u975e\u5e38\u7a33\u5065\u3002\u4f8b\u5982\uff0c\u5728\u6570\u767e\u4e07\u4e2a\u86cb\u767d\u8d28\u5e8f\u5217\u4e0a\u8bad\u7ec3\u7684 RNN \u5df2\u7ecf\u663e\u793a\u51fa\u6355\u6349\u8fdb\u5316\u548c\u7ed3\u6784\u4fe1\u606f\u7684\u80fd\u529b\uff0c\u5e76\u4e14\u53ef\u4ee5\u5e94\u7528\u4e8e\u5404\u79cd\u76d1\u7763\u4efb\u52a1\uff0c\u5305\u62ec\u4e0e\u65b0\u86cb\u767d\u8d28\u5e8f\u5217\u8bbe\u8ba1\u76f8\u5173\u7684\u4efb\u52a1 [68]\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u6ce8\u610f\u529b\u673a\u5236\u7684\u4f5c\u7528\u548cTransformer\u7684\u4f7f\u7528<\/span><\/strong>\u3002RNN \u53d1\u73b0\u7684\u4e00\u4e2a\u95ee\u9898\u662f\u5b83\u4eec\u5728\u68c0\u67e5\u8f93\u5165\u5e8f\u5217\u7684\u7279\u5b9a\u90e8\u5206<\/span>\uff08\u5bf9\u4e8e\u751f\u6210\u9ad8\u5ea6\u51c6\u786e\u7684\u8f93\u51fa\u5f88\u91cd\u8981\uff09<\/span>\u65f6\u9047\u5230\u56f0\u96be\u3002\u5411 RNN \u6dfb\u52a0\u6ce8\u610f\u529b\u673a\u5236<\/span>\uff08attention mechanism\uff09<\/span>\uff0c\u5141\u8bb8\u6a21\u578b\u5728\u8ba1\u7b97\u6bcf\u4e2a\u8f93\u51fa\u65f6\u8bbf\u95ee\u8f93\u5165\u5e8f\u5217\u7684\u6240\u6709\u90e8\u5206\uff0c\u4ee5\u7f13\u89e3\u8fd9\u4e2a\u95ee\u9898\u3002\u6700\u8fd1\u7684\u7814\u7a76\u8868\u660e\uff0c\u751a\u81f3\u6839\u672c\u4e0d\u9700\u8981 RNN \uff0c\u5355\u72ec\u4f7f\u7528\u6ce8\u610f\u529b\u673a\u5236\u5c31\u53ef\u4ee5\u4e86\u2014\u2014\u7531\u6b64\u4ea7\u751f\u7684\u6a21\u578b\u88ab\u79f0\u4e3a\u201c\u8f6c\u6362\u5668\u201d<\/span>\uff08transformer\uff09<\/span>\uff0c\u5728\u8bb8\u591a\u81ea\u7136\u8bed\u8a00\u5904\u7406\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u83b7\u5f97\u4e86\u4f18\u79c0\u7684\u9884\u6d4b\u7ed3\u679c[69]\u3002Transformer \u6a21\u578b\u6700\u8fd1\u5728\u751f\u7269\u5e8f\u5217\u4efb\u52a1\u4e0a\u8868\u73b0\u51fa\u6bd4 RNN \u66f4\u9ad8\u7684\u51c6\u786e\u6027\uff0c\u4f46\u8fd9\u4e9b\u65b9\u6cd5<\/span>\uff08\u901a\u5e38\u4f7f\u7528\u6570\u5343\u4e2a\u56fe\u5f62\u5904\u7406\u5355\u5143\uff08GPU\uff09\u5bf9\u6570\u5341\u4ebf\u4e2a\u5e8f\u5217\u8fdb\u884c\u8bad\u7ec3\uff09<\/span>\u662f\u5426\u80fd\u591f\u80dc\u8fc7\u73b0\u6709\u7684\u57fa\u4e8e\u5bf9\u9f50\u7684\u65b9\u5f0f\uff0c\u8fd8\u6709\u5f85\u89c2\u5bdf\u3002\u57fa\u4e8e\u751f\u7269\u4fe1\u606f\u5b66\u4e2d\u5e8f\u5217\u5206\u6790\u7684\u65b9\u6cd5[70]\u3002\u572814 \u5c4a\u86cb\u767d\u8d28\u7ed3\u6784\u9884\u6d4b\u5173\u952e\u8bc4\u4f30<\/span>\uff08CASP14\uff09<\/span>\u5b9e\u9a8c\u4e2d\uff0cAlphaFold2 \u53d6\u5f97\u4e86\u5de8\u5927\u6210\u529f\u3002\u7531\u4e8e\u8fd9\u4e2a\u5bf9\u4ece\u5e8f\u5217\u9884\u6d4b\u86cb\u767d\u8d28\u7ed3\u6784\u7684\u8ba1\u7b97\u65b9\u6cd5\u7684\u8bc4\u4f30\u662f\u76f2\u7684\uff0c\u8be5\u7ed3\u679c\u8868\u660e\u4f7f\u7528\u6ce8\u610f\u529b\u7684\u6a21\u578b\u4e5f\u6709\u671b\u7528\u4e8e\u7ed3\u6784\u751f\u7269\u5b66\u4e2d\u7684\u4efb\u52a1 [71] \u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u56fe\u5377\u79ef\u7f51\u7edc<\/span><\/strong>\u3002\u56fe\u5377\u79ef\u7f51\u7edc<\/span>\uff08graph convolutional networks\uff09<\/span>\u7279\u522b\u9002\u7528\u8fd9\u7c7b\u6570\u636e\u2014\u2014\u867d\u7136\u6ca1\u6709\u4efb\u4f55\u660e\u663e\u7684\u53ef\u89c1\u7ed3\u6784<\/span>\uff08\u5982\u56fe\u50cf\uff09<\/span>\uff0c\u4f46\u4ecd\u7136\u7531\u901a\u8fc7\u4efb\u610f\u6307\u5b9a\u5173\u7cfb\u6216\u4ea4\u4e92\u8fde\u63a5\u7684\u5b9e\u4f53\u7ec4\u6210 [72]\u3002\u4e0e\u751f\u7269\u5b66\u76f8\u5173\u7684\u6b64\u7c7b\u6570\u636e\u7684\u4f8b\u5b50\u5305\u62ec\u5206\u5b50<\/span>\uff08\u7531\u539f\u5b50\u548c\u952e\u7ec4\u6210\uff09<\/span>[73-76] \u548c\u86cb\u767d\u8d28-\u86cb\u767d\u8d28\u76f8\u4e92\u4f5c\u7528\u7f51\u7edc<\/span>\uff08\u7531\u86cb\u767d\u8d28\u548c\u76f8\u4e92\u4f5c\u7528\u7ec4\u6210\uff09<\/span>[77]\u3002\u5728\u8ba1\u7b97\u672f\u8bed\u4e2d\uff0c\u56fe<\/span>\uff08graph\uff09<\/span>\u53ea\u662f\u6b64\u7c7b\u6570\u636e\u7684\u8868\u793a\uff0c\u6bcf\u4e2a\u56fe\u90fd\u6709\u4e00\u7ec4\u9876\u70b9<\/span>\uff08vertex\uff09<\/span>\u6216\u8282\u70b9<\/span>\uff08node\uff09<\/span>\uff0c\u4ee5\u53ca\u4e00\u7ec4\u8868\u793a\u8282\u70b9\u4e4b\u95f4\u5404\u79cd\u7c7b\u578b\u5173\u7cfb\u6216\u8fde\u63a5\u7684\u8fb9<\/span>\uff08edge\uff09<\/span>\u3002\u7528\u4e0a\u9762\u7ed9\u51fa\u7684\u4f8b\u5b50\u6765\u89e3\u91ca\u7684\u8bdd\uff0c\u539f\u5b50\u6216\u86cb\u767d\u8d28\u7684\u8868\u793a\u53ef\u80fd\u88ab\u5f52\u7c7b\u4e3a\u8282\u70b9\u7279\u5f81<\/span>\uff08node feature\uff09<\/span>\uff0c\u800c\u952e\u6216\u76f8\u4e92\u4f5c\u7528\u53ef\u80fd\u88ab\u5f52\u7c7b\u4e3a\u8fb9\u7279\u5f81<\/span>\uff08edge feature\uff09<\/span>\u3002\u56fe\u5377\u79ef\u7f51\u7edc\u4f7f\u7528\u4ee5\u6b64\u4ea7\u751f\u7684\u56fe\u7ed3\u6784\u6765\u786e\u5b9a\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u4e2d\u7684\u4fe1\u606f\u6d41\u3002\u5982\u56fe4d\u6240\u793a\uff0c\u5f53\u6bcf\u4e2a\u8282\u70b9\u5728\u6574\u4e2a\u7f51\u7edc\u4e2d\u66f4\u65b0\u7279\u5f81\u65f6\uff0c\u4f1a\u8003\u8651\u76f8\u90bb\u8282\u70b9<\/span>\uff08adjacent node\uff09<\/span>\uff0c\u6700\u540e\u4e00\u5c42\u7684\u8282\u70b9\u7279\u5f81\u88ab\u7528\u4f5c\u8f93\u51fa<\/span>\uff08\u4f8b\u5982\uff0c\u86cb\u767d\u8d28\u4e0a\u7684\u76f8\u4e92\u4f5c\u7528\u6b8b\u57fa\uff09<\/span>\u6216\u7ec4\u5408\u5f62\u6210\u6574\u4e2a\u8f93\u51fa\u56fe<\/span>\uff08\u4f8b\u5982\uff0c\u86cb\u767d\u8d28\u7684\u6298\u53e0\u7c7b\u578b\uff09<\/span>\u3002\u4ee3\u8868\u7740\u4e0d\u540c\u5173\u8054\u7684\u56fe\u53ef\u4ee5\u5728\u8fdb\u884c\u9884\u6d4b\u65f6\u7ed3\u5408\u4e0d\u540c\u7684\u4fe1\u606f\u6765\u6e90\uff0c\u4f8b\u5982\u7ed3\u5408\u836f\u7269-\u57fa\u56e0\u548c\u98df\u7269-\u57fa\u56e0\u5173\u7cfb\u56fe\u6765\u9884\u6d4b\u9884\u9632\u764c\u75c7\u7684\u98df\u7269[78]\u3002\u7528\u4e8e\u8bad\u7ec3\u56fe\u5377\u79ef\u7f51\u7edc\u7684\u8f6f\u4ef6\u5305\u62ec PyTorch Geometric[79] \u548c Graph Nets[72]\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u81ea\u7f16\u7801\u5668<\/span><\/strong>\uff08autoencoder\uff09<\/span>\u3002\u987e\u540d\u601d\u4e49\uff0c\u81ea\u7f16\u7801\u5668\u662f\u4e00\u79cd\u795e\u7ecf\u7f51\u7edc\u67b6\u6784\uff0c\u901a\u8fc7\u5c06\u6570\u636e\u70b9\u8868\u793a\u4e3a\u5728\u5177\u6709\u9884\u5b9a\u7ef4\u5ea6<\/span>\uff08\u901a\u5e38\u8fdc\u5c0f\u4e8e\u8f93\u5165\u7ef4\u5ea6\uff09<\/span>\u7684\u65b0\u7a7a\u95f4\u4e2d\u7684\u70b9\uff0c\u6765\u5bf9\u6570\u636e\u70b9\u96c6\u5408\u8fdb\u884c\u81ea\u7f16\u7801<\/span>\uff08autoencode\uff09<\/span>\u3002\u8bad\u7ec3\u4e00\u4e2a\u795e\u7ecf\u7f51\u7edc<\/span>\uff08\u7f16\u7801\u5668\uff0cencoder\uff09<\/span>\u5c06\u8f93\u5165\u8f6c\u6362\u4e3a\u7d27\u51d1\u7684\u5185\u90e8\u8868\u793a\uff0c\u79f0\u4e3a\u201c\u6f5c\u5728\u5411\u91cf\u201d<\/span>\uff08latent vector\uff09<\/span>\u6216\u201c\u6f5c\u5728\u8868\u793a\u201d<\/span>\uff08latent representation\uff09<\/span>\uff0c\u8868\u793a\u65b0\u7a7a\u95f4\u4e2d\u7684\u5355\u4e2a\u70b9\u3002\u81ea\u7f16\u7801\u5668\u7684\u7b2c\u4e8c\u90e8\u5206\u79f0\u4e3a\u201c\u89e3\u7801\u5668\u201d<\/span>\uff08decoder\uff09<\/span>\uff0c\u5c06\u6f5c\u5728\u5411\u91cf\u4f5c\u4e3a\u8f93\u5165\uff0c\u7ecf\u8fc7\u8bad\u7ec3\u4ea7\u751f\u5177\u6709\u539f\u59cb\u7ef4\u5ea6\u7684\u539f\u59cb\u6570\u636e\uff0c\u4f5c\u4e3a\u8f93\u51fa<\/span>\uff08\u56fe 4e\uff09<\/span>\u3002\u53e6\u4e00\u79cd\u770b\u5f85\u8fd9\u4e2a\u95ee\u9898\u7684\u65b9\u6cd5\u662f\u7f16\u7801\u5668\u5c1d\u8bd5\u538b\u7f29\u8f93\u5165\uff0c\u800c\u89e3\u7801\u5668\u5c1d\u8bd5\u89e3\u538b\u7f29\u3002\u7f16\u7801\u5668\u3001\u6f5c\u5728\u8868\u793a\u548c\u89e3\u7801\u5668\u540c\u65f6\u8bad\u7ec3\u3002\u5c3d\u7ba1\u8fd9\u79cd\u201c\u8f93\u51fa\u6a21\u4eff\u8f93\u5165\u201d\u7684\u505a\u6cd5\u542c\u8d77\u6765\u4f3c\u4e4e\u6beb\u65e0\u610f\u4e49\uff0c\u4f46\u5176\u76ee\u7684\u662f\u5b66\u4e60\u8f93\u5165\u6570\u636e\u7684\u65b0\u8868\u793a\uff0c\u4f7f\u5f97\u8be5\u8868\u793a\u80fd\u7d27\u51d1\u5730\u7f16\u7801\u6240\u9700\u7684\u7279\u5f81\uff0c\u4f8b\u5982\u6570\u636e\u70b9\u4e4b\u95f4\u7684\u76f8\u4f3c\u6027\uff0c\u540c\u65f6\u4ecd\u7136\u4fdd\u7559\u91cd\u5efa\u539f\u59cb\u6570\u636e\u7684\u80fd\u529b\u3002\u5e94\u7528\u5305\u62ec\u9884\u6d4b\u4e24\u4e2a\u6570\u636e\u70b9\u7684\u5173\u8054\u7a0b\u5ea6\uff0c\u4ee5\u53ca\u5f3a\u884c\u8ba9\u6570\u636e\u5728\u6f5c\u5728\u7a7a\u95f4\u4e0a\u5f62\u6210\u4e00\u4e9b\u5bf9\u8fdb\u4e00\u6b65\u9884\u6d4b\u4efb\u52a1\u6709\u7528\u7684\u7ed3\u6784\u3002\u7f16\u7801\u5668-\u89e3\u7801\u5668\u67b6\u6784\u7684\u53e6\u4e00\u4e2a\u597d\u5904\u662f\uff0c\u4e00\u65e6\u7ecf\u8fc7\u8bad\u7ec3\uff0c\u5c31\u53ef\u4ee5\u5355\u72ec\u4f7f\u7528\u89e3\u7801\u5668\u6765\u751f\u6210<\/span>\uff08\u9884\u6d4b\uff09<\/span>\u65b0\u7684\u5408\u6210\u6570\u636e\u6837\u672c\u3002\u8fd9\u4e9b\u6837\u672c\u53ef\u4ee5\u5728\u5b9e\u9a8c\u5ba4\u4e2d\u8fdb\u884c\u6d4b\u8bd5\uff0c\u6709\u52a9\u4e8e\u5408\u6210\u751f\u7269\u5b66\u7684\u7814\u7a76 [80] \u3002\u81ea\u7f16\u7801\u5668\u5df2\u5e94\u7528\u4e8e\u4e00\u7cfb\u5217\u751f\u7269\u5b66\u95ee\u9898\uff0c\u5305\u62ec\u9884\u6d4b DNA \u7532\u57fa\u5316\u72b6\u6001 [81]\u3001\u57fa\u56e0\u548c\u86cb\u767d\u8d28\u5e8f\u5217\u5de5\u7a0b [82\u300183] \u4ee5\u53ca\u5355\u7ec6\u80de RNA \u6d4b\u5e8f\u5206\u6790 [84]\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u8bad\u7ec3\u548c\u6539\u8fdb\u795e\u7ecf\u7f51\u7edc<\/span><\/strong><\/span>\u3002\u65b9\u68461 \u6982\u8ff0\u4e86\u8bad\u7ec3\u673a\u5668\u5b66\u4e60\u6a21\u578b\u7684\u4e00\u822c\u7a0b\u5e8f\u3002\u4f46\u662f\uff0c\u7531\u4e8e\u795e\u7ecf\u7f51\u7edc\u5728\u7ed3\u6784\u4e0a\u6bd4\u4f20\u7edf\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u590d\u6742\u5f97\u591a\uff0c\u56e0\u6b64\u5b58\u5728\u4e00\u4e9b\u795e\u7ecf\u7f51\u7edc\u7279\u6709\u7684\u95ee\u9898\u3002\u5728\u9009\u62e9\u4e86\u4e00\u4e2a\u795e\u7ecf\u7f51\u7edc\u4f5c\u4e3a\u9002\u5408\u9884\u671f\u5e94\u7528\u7684\u6a21\u578b<\/span>\uff08\u56fe1\uff09<\/span>\u540e\uff0c\u5148\u53ea\u5728\u5355\u4e2a\u8bad\u7ec3\u793a\u4f8b<\/span>\uff08\u4f8b\u5982\uff0c\u5355\u4e2a\u56fe\u50cf\u6216\u57fa\u56e0\u5e8f\u5217\uff09<\/span>\u4e0a\u5bf9\u5176\u8fdb\u884c\u8bad\u7ec3\uff0c\u901a\u5e38\u662f\u4e2a\u597d\u4e3b\u610f\u3002\u8fd9\u79cd\u7ecf\u8fc7\u8bad\u7ec3\u7684\u6a21\u578b\u5bf9\u4e8e\u8fdb\u884c\u9884\u6d4b\u6ca1\u6709\u7528\u5904\uff0c\u4f46\u8bad\u7ec3\u53ef\u4ee5\u5f88\u597d\u5730\u63ed\u793a\u7f16\u7a0b\u9519\u8bef\u3002\u8bad\u7ec3\u635f\u5931\u51fd\u6570\u5e94\u8be5\u5f88\u5feb\u53d8\u4e3a\u96f6\uff0c\u56e0\u4e3a\u7f51\u7edc\u53ea\u662f\u7b80\u5355\u5730\u8bb0\u4f4f\u8f93\u5165\uff1b\u5982\u679c\u4e0d\u662f\uff0c\u5219\u4ee3\u7801\u4e2d\u53ef\u80fd\u5b58\u5728\u9519\u8bef\uff0c\u6216\u8005\u7b97\u6cd5\u4e0d\u591f\u590d\u6742\uff0c\u65e0\u6cd5\u5bf9\u8f93\u5165\u6570\u636e\u8fdb\u884c\u5efa\u6a21\u3002\u4e00\u65e6\u7f51\u7edc\u901a\u8fc7\u4e86\u8fd9\u4e2a\u57fa\u672c\u7684\u8c03\u8bd5\u6d4b\u8bd5\uff0c\u5c31\u53ef\u4ee5\u7ee7\u7eed\u5bf9\u6574\u4e2a\u8bad\u7ec3\u96c6\u8fdb\u884c\u8bad\u7ec3\uff0c\u6765\u6700\u5c0f\u5316\u8bad\u7ec3\u635f\u5931\u51fd\u6570\u3002\u8fd9\u53ef\u80fd\u9700\u8981\u8c03\u6574\u5b66\u4e60\u7387\u7b49\u8d85\u53c2\u6570<\/span>\uff08\u56fe 2e\uff09<\/span>\u3002\u901a\u8fc7\u76d1\u63a7\u8bad\u7ec3\u96c6\u548c\u9a8c\u8bc1\u96c6\u7684\u635f\u5931\uff0c\u53ef\u4ee5\u68c0\u6d4b\u5230\u7f51\u7edc\u7684\u8fc7\u5ea6\u62df\u5408\u2014\u2014\u8bad\u7ec3\u635f\u5931\u7ee7\u7eed\u964d\u4f4e\uff0c\u9a8c\u8bc1\u96c6\u7684\u635f\u5931\u5f00\u59cb\u589e\u52a0\u3002\u8bad\u7ec3\u901a\u5e38\u5728\u90a3\u4e2a\u65f6\u5019\u505c\u6b62\uff0c\u8fd9\u4e2a\u8fc7\u7a0b\u88ab\u79f0\u4e3a\u63d0\u524d\u505c\u6b62<\/span>\uff08\u56fe 2f\uff09<\/span>\u3002\u795e\u7ecf\u7f51\u7edc<\/span>\uff08\u6216\u4efb\u4f55\u673a\u5668\u5b66\u4e60\u6a21\u578b\uff09<\/span>\u7684\u8fc7\u5ea6\u62df\u5408\uff0c\u5982\u56fe 1d\u6240\u793a\uff0c\u610f\u5473\u7740\u6a21\u578b\u5f00\u59cb\u7b80\u5355\u5730\u8bb0\u4f4f\u8bad\u7ec3\u96c6\u7684\u7279\u5f81\uff0c\u56e0\u6b64\u5f00\u59cb\u5931\u53bb\u6cdb\u5316\u5230\u65b0\u6570\u636e\u7684\u80fd\u529b\u3002\u63d0\u524d\u505c\u6b62\u662f\u9632\u6b62\u8fd9\u79cd\u60c5\u51b5\u7684\u597d\u65b9\u6cd5\uff0c\u4f46\u5728\u8bad\u7ec3\u671f\u95f4\u53ef\u4ee5\u4f7f\u7528\u5176\u4ed6\u6280\u672f\uff0c\u4f8b\u5982\u6a21\u578b\u7684\u6b63\u5219\u5316<\/span>\uff08regularization\uff09<\/span>\u6216\u4e22\u5f03<\/span>\uff08dropout\uff09<\/span>\u6280\u672f\u3002\u4e22\u5f03\u6280\u672f\u968f\u673a\u5ffd\u7565\u7f51\u7edc\u4e2d\u7684\u4e00\u4e9b\u8282\u70b9\uff0c\u6765\u8feb\u4f7f\u7f51\u7edc\u5b66\u4e60\u6d89\u53ca\u591a\u4e2a\u8282\u70b9\u7684\u66f4\u5f3a\u5927\u7684\u9884\u6d4b\u7b56\u7565\u3002<\/span><\/section>\n
        <\/span><\/section>\n
\u7528\u4e8e\u8bad\u7ec3\u795e\u7ecf\u7f51\u7edc\u7684\u6d41\u884c\u8f6f\u4ef6\u5305\u5305\u62ec PyTorch [85] \u548c Tensorflow [86]\u3002\u8bad\u7ec3\u795e\u7ecf\u7f51\u7edc\u7684\u8ba1\u7b97\u8981\u6c42\u5f88\u9ad8\uff0c\u901a\u5e38\u9700\u8981\u5177\u6709\u8db3\u591f\u5185\u5b58\u7684\u56fe\u5f62\u5904\u7406\u5355\u5143\u6216\u5f20\u91cf\u5904\u7406\u5355\u5143\uff0c\u56e0\u4e3a\u8fd9\u4e9b\u8bbe\u5907\u53ef\u4ee5\u63d0\u4f9b\u6bd4\u4f7f\u7528\u6807\u51c6\u4e2d\u592e\u5904\u7406\u5355\u5143 10 \u5230 100 \u500d\u7684\u52a0\u901f\u3002\u5728\u8bad\u7ec3\u8fd1\u5e74\u6765\u53d6\u5f97\u6210\u529f\u7684\u5927\u578b\u6a21\u578b\u4ee5\u53ca\u5728\u5927\u578b\u6570\u636e\u96c6\u4e0a\u8fdb\u884c\u8bad\u7ec3\u65f6\uff0c\u9700\u8981\u8fd9\u79cd\u52a0\u901f\u3002\u7136\u800c\uff0c\u8fd0\u884c\u4e00\u4e2a\u5df2\u7ecf\u8bad\u7ec3\u597d\u7684\u6a21\u578b\u901a\u5e38\u8981\u5feb\u5f97\u591a\uff0c\u800c\u4e14\u901a\u5e38\u53ea\u5728\u666e\u901a\u7684\u4e2d\u592e\u5904\u7406\u5355\u5143\u4e0a\u662f\u53ef\u884c\u7684\u3002\u90a3\u4e9b\u65e0\u6cd5\u8bbf\u95ee\u56fe\u5f62\u5904\u7406\u5355\u5143\u8fdb\u884c\u8bad\u7ec3\u7684\u4eba\uff0c\u4e5f\u53ef\u4ee5\u4f7f\u7528\u901a\u7528\u63d0\u4f9b\u5546\u7684\u4e91\u8ba1\u7b97\u89e3\u51b3\u65b9\u6848\u3002\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0c\u5bf9\u4e8e\u5c0f\u4efb\u52a1\uff0cColaboratory <\/span>(Colab) <\/span>\u5141\u8bb8\u5728\u56fe\u5f62\u5904\u7406\u5355\u5143\u6216\u5f20\u91cf\u5904\u7406\u5355\u5143\u4e0a\u514d\u8d39\u6d4b\u8bd5 Python \u4ee3\u7801\u3002\u4f7f\u7528 Colab \u662f\u5f00\u59cb\u4f7f\u7528\u57fa\u4e8e Python \u7684\u6df1\u5ea6\u5b66\u4e60\u7684\u7edd\u4f73\u65b9\u5f0f\u3002<\/span><\/section>\n
 <\/span><\/section>\n

\u56fe4. \u795e\u7ecf\u7f51\u7edc\u65b9\u6cd5\u3002(a) <\/strong>\u591a\u5c42\u611f\u77e5\u673a\u7531\u8282\u70b9\uff08\u663e\u793a\u4e3a\u5706\uff09\u7ec4\u6210\uff0c\u4ee3\u8868\u6570\u5b57\uff1a\u8f93\u5165\u503c\u3001\u8f93\u51fa\u503c\u6216\u5185\u90e8\uff08\u9690\u85cf\uff09\u503c\u3002\u8282\u70b9\u6392\u5217\u5728\u5177\u6709\u8fde\u63a5\u7684\u5c42\u4e2d\uff0c\u8fd9\u4e9b\u8fde\u63a5\u610f\u5473\u7740\u5b66\u4e60\u540e\u7684\u53c2\u6570\uff0c\u4f4d\u4e8e\u4e00\u5c42\u7684\u6bcf\u4e2a\u8282\u70b9\u548c\u4e0b\u4e00\u5c42\u7684\u6bcf\u4e2a\u8282\u70b9\u4e4b\u95f4\u3002\u4f8b\u5982\uff0c\u5206\u5b50\u7279\u6027\u53ef\u7528\u4e8e\u9884\u6d4b\u836f\u7269\u6bd2\u6027\uff0c\u56e0\u4e3a\u9884\u6d4b\u53ef\u4ee5\u4ece\u72ec\u7acb\u8f93\u5165\u7279\u5f81\uff08\u5206\u5b50\u7279\u6027\uff09\u7684\u4e00\u4e9b\u590d\u6742\u7ec4\u5408\u4e2d\u8fdb\u884c\u3002(b)<\/strong> \u5377\u79ef\u795e\u7ecf\u7f51\u7edc (CNN) \u4f7f\u7528\u8de8\u8f93\u5165\u5c42\u79fb\u52a8\u7684\u8fc7\u6ee4\u5668\uff0c\u6765\u8ba1\u7b97\u4e0b\u4e00\u5c42\u4e2d\u7684\u503c\u3002\u8fc7\u6ee4\u5668\u8de8\u6574\u4e2a\u5c42\u8fd0\u884c\uff0c\u610f\u5473\u7740\u53c2\u6570\u662f\u5171\u4eab\u7684\uff0c\u65e0\u8bba\u4f4d\u7f6e\u5982\u4f55\uff0c\u90fd\u53ef\u4ee5\u68c0\u6d4b\u5230\u76f8\u4f3c\u7684\u5b9e\u4f53\u3002\u4e8c\u7ef4 CNN \u663e\u793a\u5728\u663e\u5fae\u955c\u56fe\u50cf\u4e0a\u8fd0\u884c\uff0c\u4f46\u4e00\u7ef4\u548c\u4e09\u7ef4 CNN \u5728\u751f\u7269\u5b66\u4e2d\u4e5f\u6709\u5e94\u7528\u3002\u8fd9\u91cc\u7684\u7ef4\u5ea6\u6307\u7684\u662f\u6570\u636e\u7684\u7a7a\u95f4\u7ef4\u5ea6\uff1b\u4e0e\u6b64\u76f8\u5bf9\u5e94\uff0c\u53ef\u4ee5\u914d\u7f6e CNN \u5185\u90e8\u7684\u8fde\u63a5\u6027\u3002\u4f8b\u5982\uff0c\u751f\u7269\u5e8f\u5217\u53ef\u4ee5\u88ab\u8ba4\u4e3a\u662f\u4e00\u7ef4\u7684\uff0c\u800c\u78c1\u5171\u632f\u6210\u50cf\u6570\u636e\u53ef\u4ee5\u88ab\u8ba4\u4e3a\u662f\u4e09\u7ef4\u7684\u3002(c) <\/strong>\u5faa\u73af\u795e\u7ecf\u7f51\u7edc (RNN) \u4f7f\u7528\u76f8\u540c\u7684\u5b66\u5230\u7684\u53c2\u6570\u6765\u5904\u7406\u987a\u5e8f\u8f93\u5165\u7684\u6bcf\u4e2a\u90e8\u5206\uff0c\u4e3a\u6bcf\u4e2a\u8f93\u5165\u63d0\u4f9b\u8f93\u51fa\u548c\u66f4\u65b0\u7684\u9690\u85cf\u72b6\u6001\u3002\u9690\u85cf\u72b6\u6001\u7528\u4e8e\u643a\u5e26\u5e8f\u5217\u524d\u9762\u90e8\u5206\u7684\u4fe1\u606f\u3002\u5728\u8fd9\u4e2a\u4f8b\u5b50\u91cc\uff0c\u53ef\u4ee5\u9884\u6d4b DNA \u5e8f\u5217\u4e2d\u6bcf\u4e2a\u78b1\u57fa\u4e0e\u8f6c\u5f55\u56e0\u5b50\u7684\u7ed3\u5408\u6982\u7387\u3002RNN \u88ab\u5c55\u5f00\uff0c\u65b9\u4fbf\u5c55\u793a\u6bcf\u4e2a\u8f93\u51fa\u662f\u5982\u4f55\u4f7f\u7528\u76f8\u540c\u7684\u5c42\u751f\u6210\u7684\uff1b\u4e0d\u5e94\u8be5\u88ab\u6df7\u6dc6\u4e3a\u662f\u4f7f\u7528\u4e0d\u540c\u7684\u5c42\u3002(d) <\/strong>\u56fe\u5377\u79ef\u7f51\u7edc\u4f7f\u7528\u6765\u81ea\u56fe\u4e2d\u8fde\u63a5\u8282\u70b9\u7684\u4fe1\u606f\uff08\u4f8b\u5982\u86cb\u767d\u8d28-\u86cb\u767d\u8d28\u76f8\u4e92\u4f5c\u7528\u7f51\u7edc\uff09\uff0c\u901a\u8fc7\u7ec4\u5408\u6765\u81ea\u6240\u6709\u76f8\u90bb\u8282\u70b9\u7684\u9884\u6d4b\uff0c\u6765\u66f4\u65b0\u7f51\u7edc\u4e2d\u7684\u8282\u70b9\u5c5e\u6027\u3002\u66f4\u65b0\u540e\u7684\u8282\u70b9\u5c5e\u6027\u5f62\u6210\u7f51\u7edc\u4e2d\u7684\u4e0b\u4e00\u5c42\uff0c\u5e76\u5728\u8f93\u51fa\u5c42\u4e2d\u9884\u6d4b\u6240\u9700\u7684\u5c5e\u6027\u3002(e) \u81ea\u7f16\u7801\u5668\u7531\u7f16\u7801\u5668\u795e\u7ecf\u7f51\u7edc\u548c\u89e3\u7801\u5668\u795e\u7ecf\u7f51\u7edc\u7ec4\u6210\uff0c\u7f16\u7801\u5668\u795e\u7ecf\u7f51\u7edc\u5c06\u8f93\u5165\u8f6c\u6362\u4e3a\u4f4e\u7ef4\u6f5c\u5728\u8868\u793a\uff0c\u89e3\u7801\u5668\u795e\u7ecf\u7f51\u7edc\u5c06\u8fd9\u79cd\u6f5c\u5728\u8868\u793a\u8f6c\u6362\u56de\u539f\u59cb\u8f93\u5165\u5f62\u5f0f\u3002\u4f8b\u5982\uff0c\u53ef\u4ee5\u7f16\u7801\u86cb\u767d\u8d28\u5e8f\u5217\u5e76\u4f7f\u7528\u6f5c\u5728\u8868\u793a\u6765\u751f\u6210\u65b0\u7684\u86cb\u767d\u8d28\u5e8f\u5217\u3002\u5728\u793a\u4f8b\u4e2d\uff0c5 \u4e2a\u6b8b\u57fa\u4e2d\u6709 4 \u4e2a\u4e0e\u81ea\u52a8\u7f16\u7801\u5668\u7f16\u7801\u548c\u89e3\u7801\u540e\u7684\u8f93\u5165\u76f8\u540c\uff0c\u8868\u660e\u5728\u8be5\u5e8f\u5217\u4e0a\u7684\u51c6\u786e\u7387\u4e3a 80%\u3002<\/span><\/p>\n

       <\/span><\/section>\n


<\/p>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

\u751f\u7269\u5e94\u7528\u7684\u6311\u6218<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n


<\/p>\n

\u751f\u7269\u6570\u636e\u5efa\u6a21\u7684\u6700\u5927\u6311\u6218\u4e5f\u8bb8\u662f\u79cd\u7c7b\u7e41\u591a\u3002\u751f\u7269\u5b66\u5bb6\u4f7f\u7528\u7684\u6570\u636e\u5305\u62ec\u57fa\u56e0\u548c\u86cb\u767d\u8d28\u5e8f\u5217\u3001\u57fa\u56e0\u968f\u65f6\u95f4\u7684\u8868\u8fbe\u6c34\u5e73\u3001\u8fdb\u5316\u6811\u3001\u663e\u5fae\u955c\u56fe\u50cf\u30013D \u7ed3\u6784\u548c\u76f8\u4e92\u4f5c\u7528\u7f51\u7edc\u7b49\u3002\u88682 \u603b\u7ed3\u4e86\u9488\u5bf9\u7279\u5b9a\u751f\u7269\u6570\u636e\u7c7b\u578b\u7684\u4e00\u4e9b\u6700\u4f73\u505a\u6cd5\u548c\u91cd\u8981\u8003\u8651\u56e0\u7d20\u3002\u7531\u4e8e\u9047\u5230\u7684\u6570\u636e\u7c7b\u578b\u7684\u591a\u6837\u6027\uff0c\u751f\u7269\u6570\u636e\u901a\u5e38\u9700\u8981\u4e00\u4e9b\u5b9a\u5236\u7684\u89e3\u51b3\u65b9\u6848\u6765\u6709\u6548\u5730\u5904\u7406\u3002\u8981\u5728\u8fd9\u4e9b\u95ee\u9898\u9886\u57df\u4e2d\u4f7f\u7528\u673a\u5668\u5b66\u4e60\uff0c\u5f88\u96be\u76f4\u63a5\u63a8\u8350\u5df2\u6709\u6a21\u578b\uff0c\u66f4\u4e0d\u8981\u8bf4\u901a\u7528\u6307\u5357\uff0c\u56e0\u4e3a\u6a21\u578b\u3001\u8bad\u7ec3\u7a0b\u5e8f\u548c\u6d4b\u8bd5\u6570\u636e\u7684\u9009\u62e9\u5c06\u5728\u5f88\u5927\u7a0b\u5ea6\u4e0a\u53d6\u51b3\u4e8e\u4eba\u4eec\u60f3\u8981\u56de\u7b54\u7684\u786e\u5207\u95ee\u9898\u3002\u5c3d\u7ba1\u5982\u6b64\uff0c\u8981\u6210\u529f\u4f7f\u7528\u673a\u5668\u5b66\u4e60<\/span>\uff08\u4e0d\u7ba1\u662f\u5728\u751f\u7269\u5b66\u4e2d\u8fd8\u662f\u5728\u66f4\u5e7f\u6cdb\u7684\u610f\u4e49\u4e0a\uff09<\/span>\uff0c\u8fd8\u662f\u6709\u4e00\u4e9b\u5e38\u89c1\u7684\u95ee\u9898\u9700\u8981\u8003\u8651\u3002<\/span><\/section>\n
 <\/span><\/section>\n

\n
\u88682  \u5bf9\u4e8e\u4e0d\u540c\u751f\u7269\u6570\u636e\u7c7b\u578b\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u7b56\u7565\u7684\u5efa\u8bae<\/span><\/strong><\/section>\n
<\/section>\n
<\/section>\n
<\/section>\n
 <\/span>
<\/section>\n<\/h2>\n
\u6570\u636e\u53ef\u5f97\u6027<\/span><\/strong><\/span>\u3002\u751f\u7269\u5b66\u7684\u72ec\u7279\u4e4b\u5904\u5728\u4e8e\uff0c\u6709\u4e00\u4e9b\u95ee\u9898\u9886\u57df\u7684\u516c\u5f00\u6570\u636e\u91cf\u975e\u5e38\u5927\uff0c\u800c\u5176\u5b83\u95ee\u9898\u9886\u57df\u7684\u6570\u636e\u91cf\u975e\u5e38\u5c0f\u3002\u6bd4\u5982\uff0c\u516c\u5171\u6570\u636e\u5e93<\/span>\uff08\u5982 GenBank \u548c UniProt\uff09<\/span>\u4e2d\u7684\u751f\u7269\u5e8f\u5217\u6570\u636e\u76f8\u5bf9\u4e30\u5bcc\uff0c\u800c\u5173\u4e8e\u86cb\u767d\u8d28\u76f8\u4e92\u4f5c\u7528\u7684\u53ef\u9760\u6570\u636e\u5219\u5f88\u96be\u83b7\u5f97\u3002\u53ef\u7528\u4e8e\u7ed9\u5b9a\u95ee\u9898\u7684\u6570\u636e\u91cf\uff0c\u5bf9\u53ef\u4ee5\u6709\u6548\u4f7f\u7528\u7684\u6280\u672f\u7684\u9009\u62e9\uff0c\u5177\u6709\u6df1\u8fdc\u7684\u5f71\u54cd\u3002\u4e00\u4e2a\u975e\u5e38\u7c97\u7565\u7684\u6307\u5bfc\u65b9\u9488\u662f\uff0c\u5f53\u53ea\u6709\u5c11\u91cf\u6570\u636e<\/span>\uff08\u5982\u4e09\u56db\u4f4d\u6570\uff09<\/span>\u53ef\u7528\u65f6\uff0c\u4eba\u4eec\u57fa\u672c\u4e0a\u88ab\u8feb\u4f7f\u7528\u66f4\u4f20\u7edf\u7684\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u2014\u2014\u66f4\u6709\u53ef\u80fd\u4ea7\u751f\u53ef\u9760\u7684\u9884\u6d4b\u3002\u5f53\u6709\u66f4\u591a\u7684\u6570\u91cf\u53ef\u7528\u65f6\uff0c\u7814\u7a76\u4eba\u5458\u53ef\u4ee5\u5f00\u59cb\u8003\u8651\u66f4\u591a\u53c2\u6570\u5316\u7684\u6a21\u578b\u2014\u2014\u4f8b\u5982\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\u3002\u5728\u6709\u76d1\u7763\u7684\u673a\u5668\u5b66\u4e60\u4e2d\uff0c\u8fd8\u5e94\u8be5\u8003\u8651\u6570\u636e\u96c6\u4e2d\u6bcf\u4e2a\u771f\u5b9e\u6807\u7b7e\u7684\u76f8\u5bf9\u6bd4\u4f8b\uff0c\u5982\u679c\u67d0\u4e9b\u6807\u7b7e\u5f88\u5c11\u89c1\uff0c\u673a\u5668\u5b66\u4e60\u9700\u8981\u66f4\u591a\u7684\u6570\u636e[87]\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u6570\u636e\u6cc4\u9732<\/strong><\/span>\u3002\u5c3d\u7ba1\u751f\u7269\u6570\u636e\u7684\u89c4\u6a21\u548c\u590d\u6742\u6027\u53ef\u80fd\u4f7f\u5b83\u4eec\u770b\u8d77\u6765\u975e\u5e38\u9002\u5408\u673a\u5668\u5b66\u4e60\uff0c\u4f46\u4ecd\u6709\u4e00\u4e9b\u91cd\u8981\u7684\u8003\u8651\u56e0\u7d20\u9700\u8981\u7262\u8bb0 [21, 88, 89]\u3002\u4e00\u4e2a\u5173\u952e\u95ee\u9898\u662f\u5982\u4f55\u9a8c\u8bc1\u6a21\u578b\u7684\u6027\u80fd<\/strong>\u3002\u8bad\u7ec3\u96c6\u3001\u9a8c\u8bc1\u96c6\u548c\u6d4b\u8bd5\u96c6\u7684\u5171\u540c\u8bbe\u7f6e\u53ef\u80fd\u4f1a\u5bfc\u81f4\u4e00\u4e9b\u95ee\u9898\uff0c\u4f8b\u5982\u7814\u7a76\u4eba\u5458\u4f7f\u7528\u5404\u79cd\u6a21\u578b\uff0c\u5728\u540c\u4e00\u4e2a\u6d4b\u8bd5\u96c6\u4e0a\u91cd\u590d\u6d4b\u8bd5\uff0c\u4ee5\u83b7\u5f97\u6700\u5927\u7684\u51c6\u786e\u5ea6\uff0c\u4ece\u800c\u6709\u9ad8\u4f30\u5176\u6027\u80fd\u7684\u98ce\u9669\uff0c\u65e0\u6cd5\u5c06\u5176\u63a8\u5e7f\u5230\u5176\u4ed6\u6d4b\u8bd5\u96c6\u6216\u65b0\u7684\u6d4b\u8bd5\u96c6\u3002\u7136\u800c\uff0c\u751f\u7269\u6570\u636e\u63d0\u51fa\u4e86\u4e00\u4e2a\u66f4\u91cd\u8981\u7684\u95ee\u9898\uff1a\u5728\u5177\u6709\u76f8\u5173\u6761\u76ee\u7684\u5927\u578b\u6570\u636e\u96c6<\/strong><\/span>\uff08\u4f8b\u5982\uff0c\u7531\u4e8e\u5bb6\u65cf\u5173\u7cfb\u6216\u8fdb\u5316\u5173\u7cfb\uff09<\/span>\u4e2d\uff0c\u5982\u4f55\u786e\u4fdd\u4e24\u4e2a\u5bc6\u5207\u76f8\u5173\u7684\u6761\u76ee\u6700\u7ec8\u4e0d\u4f1a\u88ab\u5206\u5230\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u91cc\u53bb\uff1f<\/strong>\u5982\u679c\u53d1\u751f\u8fd9\u79cd\u60c5\u51b5\uff0c\u5219\u7814\u7a76\u4eba\u5458\u6d4b\u8bd5\u7684\uff0c\u662f\u6a21\u578b\u8bb0\u4f4f\u7279\u5b9a\u6848\u4f8b\u7684\u80fd\u529b\uff0c\u800c\u4e0d\u662f\u5176\u9884\u6d4b\u76f8\u5173\u5c5e\u6027\u7684\u80fd\u529b\u3002\u8fd9\u662f\u901a\u5e38\u88ab\u79f0\u4e3a\u201c\u6570\u636e\u6cc4\u6f0f\u201d<\/span>\uff08Data Leakage\uff09<\/span>\u7684\u95ee\u9898\u7684\u4e00\u4e2a\u4f8b\u5b50\uff0c\u5e76\u5bfc\u81f4\u7ed3\u679c\u770b\u8d77\u6765\u6bd4\u5b9e\u9645\u60c5\u51b5\u597d\uff0c\u8fd9\u53ef\u80fd\u662f\u7814\u7a76\u4eba\u5458\u4e0d\u613f\u610f\u4e25\u683c\u5bf9\u5f85\u8fd9\u4e2a\u95ee\u9898\u7684\u539f\u56e0\u8d28\u7591\u3002\u5176\u5b83\u7c7b\u578b\u7684\u6570\u636e\u6cc4\u6f0f\u4e5f\u662f\u53ef\u80fd\u7684<\/span>\uff08\u4f8b\u5982\uff0c\u5728\u8bad\u7ec3\u671f\u95f4\u4f7f\u7528\u4efb\u4f55\u5728\u6d4b\u8bd5\u671f\u95f4\u4e0d\u53ef\u7528\u7684\u6570\u636e\u6216\u7279\u5f81\uff09<\/span>\u3002\u63a5\u4e0b\u6765\u7684\u8ba8\u8bba\uff0c\u4e3b\u8981\u5173\u6ce8\u4e8e\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u4e2d\u6837\u672c\u662f\u5426\u76f8\u5173\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u8fd9\u91cc\u6240\u8bf4\u7684\u201c\u76f8\u5173\u201d\u662f\u4ec0\u4e48\u610f\u601d\uff0c\u53d6\u51b3\u4e8e\u7814\u7a76\u7684\u6027\u8d28\u3002\u53ef\u80fd\u662f\u4ece\u540c\u4e00\u60a3\u8005\u6216\u540c\u4e00\u751f\u7269\u4f53\u91c7\u6837\u6570\u636e\u7684\u60c5\u51b5\u3002\u7136\u800c\uff0c\u751f\u7269\u5b66\u4e2d\u53d1\u751f\u6570\u636e\u6cc4\u6f0f\u7684\u7ecf\u5178\u60c5\u51b5\uff0c\u51fa\u73b0\u5728\u5bf9\u86cb\u767d\u8d28\u5e8f\u5217\u548c\u7ed3\u6784\u7684\u7814\u7a76\u4e2d\u3002\u6bd4\u8f83\u5e38\u89c1\u4f46\u901a\u5e38\u9519\u8bef\u7684\u662f\uff0c\u7814\u7a76\u4eba\u5458\u8bd5\u56fe\u786e\u4fdd\u8bad\u7ec3\u96c6\u4e2d\u7684\u86cb\u767d\u8d28\u4e0e\u6d4b\u8bd5\u96c6\u4e2d\u7684\u4efb\u4f55\u86cb\u767d\u8d28\u7684\u5e8f\u5217\u540c\u4e00\u6027\u90fd\u4e0d\u4f1a\u8d85\u8fc7\u67d0\u4e2a\u9608\u503c\uff0c\u901a\u5e38\u9608\u503c\u4e3a 30% \u6216 25%\u3002\u8fd9\u8db3\u4ee5\u6392\u9664\u8bb8\u591a\u540c\u6e90\u86cb\u767d\u8d28\u5bf9\uff0c\u4f46\u4eba\u4eec\u5df2\u7ecf\u77e5\u9053\u51e0\u5341\u5e74\u4e86\u7684\u662f\uff0c\u4e00\u4e9b\u540c\u6e90\u86cb\u767d\u8d28\u51e0\u4e4e\u6ca1\u6709\u5e8f\u5217\u76f8\u4f3c\u6027 [90, 91]\uff0c\u8fd9\u610f\u5473\u7740\u7b80\u5355\u5730\u6309\u5e8f\u5217\u540c\u4e00\u6027\u8fc7\u6ee4\uff0c\u4e0d\u8db3\u4ee5\u9632\u6b62\u6570\u636e\u6cc4\u6f0f\u3002\u8fd9\u5bf9\u4e8e\u5c06\u5e8f\u5217\u6bd4\u5bf9\u6216\u5e8f\u5217\u914d\u7f6e\u6587\u4ef6\u4f5c\u4e3a\u8f93\u5165\u8fdb\u884c\u64cd\u4f5c\u7684\u6a21\u578b\u5c24\u4e3a\u91cd\u8981\uff0c\u56e0\u4e3a\u5c3d\u7ba1\u4e24\u4e2a\u5355\u72ec\u7684\u86cb\u767d\u8d28\u5e8f\u5217\u53ef\u80fd\u6ca1\u6709\u4efb\u4f55\u660e\u663e\u7684\u76f8\u4f3c\u6027\uff0c\u4f46\u5b83\u4eec\u7684\u751f\u7269\u5b66\u529f\u80fd\u5b9e\u9645\u4e0a\u53ef\u80fd\u76f8\u540c\u3002\u8fd9\u610f\u5473\u7740\u5bf9\u4e8e\u673a\u5668\u5b66\u4e60\u6a21\u578b\uff0c\u8fd9\u4e24\u4e2a\u6807\u7b7e\u672c\u8d28\u4e0a\u662f\u76f8\u540c\u7684\u6570\u636e\u70b9\u2014\u2014\u90fd\u5728\u63cf\u8ff0\u76f8\u540c\u7684\u86cb\u767d\u8d28\u5bb6\u65cf\u3002\u5bf9\u4e8e\u86cb\u767d\u8d28\u5e8f\u5217\uff0c\u907f\u514d\u6b64\u95ee\u9898\u7684\u4e00\u79cd\u89e3\u51b3\u65b9\u6848\u662f\u4f7f\u7528\u654f\u611f\u7684\u9690\u9a6c\u5c14\u53ef\u592b\u6a21\u578b\u8f6e\u5ed3\u6bd4\u8f83\u5de5\u5177<\/span>\uff08\u4f8b\u5982 HH-suite\uff09<\/span>\u641c\u7d22\u6d4b\u8bd5\u6570\u636e\uff0c\u8be5\u5de5\u5177\u53ef\u4ee5\u627e\u5230\u4e0e\u8bad\u7ec3\u6570\u636e[92] \u8fdc\u76f8\u5173\u7684\u5e8f\u5217\u3002\u5728\u5e38\u89c1\u7684\u86cb\u767d\u8d28\u7ed3\u6784\u88ab\u7528\u4f5c\u8f93\u5165\u6216\u8f93\u51fa\u7684\u60c5\u51b5\u4e0b\uff0c\u7ed3\u6784\u5206\u7c7b\u5982 CATH [93] \u6216 ECOD [94] \u53ef\u7528\u4e8e\u6392\u9664\u76f8\u4f3c\u7684\u6298\u53e0\u6216\u540c\u6e90\u86cb\u767d\u8d28\u3002\u7c7b\u4f3c\u7684\u95ee\u9898\u4f1a\u5f71\u54cd\u9884\u6d4b\u86cb\u767d\u8d28-\u914d\u4f53\u7ed3\u5408\u4eb2\u548c\u529b\u7684\u7814\u7a76 [95]\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u9700\u8981\u660e\u786e\u7684\u662f\uff0c\u6570\u636e\u6cc4\u6f0f\u4e0d\u662f\u4efb\u4f55\u7279\u5b9a\u7c7b\u578b\u6570\u636e\u7684\u56fa\u6709\u95ee\u9898\uff0c\u800c\u662f\u5728\u8bad\u7ec3\u548c\u8bc4\u4f30\u673a\u5668\u5b66\u4e60\u6a21\u578b\u65f6\u5982\u4f55\u4f7f\u7528\u6570\u636e\u7684\u95ee\u9898\u3002\u7814\u7a76\u4eba\u5458\u671f\u671b\u7ecf\u8fc7\u8bad\u7ec3\u7684\u6a21\u578b\u80fd\u591f\u5728\u4e0e\u8bad\u7ec3\u96c6\u76f8\u4f3c\u7684\u6570\u636e\u4e0a\u4ea7\u751f\u975e\u5e38\u597d\u7684\u7ed3\u679c\u3002\u5f53\u5728\u67d0\u4e9b\u57fa\u51c6\u96c6\u4e0a\u770b\u8d77\u6765\u51c6\u786e\u7684\u6a21\u578b\uff0c\u5728\u5b9e\u9645\u4e0a\u4e0e\u8bad\u7ec3\u96c6\u4e0d\u540c\u7684\u65b0\u6570\u636e\u4e0a\u8868\u73b0\u4e0d\u4f73\u65f6\uff0c\u5c31\u53ef\u80fd\u53d1\u751f\u4e86\u6570\u636e\u6cc4\u6f0f\u95ee\u9898<\/strong>\uff1b\u6362\u53e5\u8bdd\u8bf4\uff0c\u8be5\u6a21\u578b\u6ca1\u6709\u6cdb\u5316\uff0c\u53ef\u80fd\u662f\u56e0\u4e3a\u5b83\u6ca1\u6709\u5bf9\u53d8\u91cf\u4e4b\u95f4\u7684\u771f\u5b9e\u5173\u7cfb\u8fdb\u884c\u5efa\u6a21\uff0c\u800c\u662f\u8bb0\u4f4f\u4e86\u6570\u636e\u4e2d\u5b58\u5728\u7684\u9690\u85cf\u5173\u8054\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u7531\u4e8e\u5ba1\u7a3f\u4eba\u7684\u4e25\u683c\u8981\u6c42\uff0c\u4e00\u4e9b\u5b66\u672f\u671f\u520a\u73b0\u5728\u5f00\u59cb\u8981\u6c42\u5728\u8003\u8651\u53d1\u8868\u8bba\u6587\u4e4b\u524d\u8fdb\u884c\u4e25\u683c\u7684\u57fa\u51c6\u6d4b\u8bd5\u3002\u5982\u679c\u6ca1\u6709\u9002\u5f53\u7684\u6d4b\u8bd5\uff0c\u6a21\u578b\u7684\u6027\u80fd\u5f88\u53ef\u80fd\u65e0\u6cd5\u4ee3\u8868\u672a\u89c1\u6570\u636e\u7684\u771f\u5b9e\u6027\u80fd\uff0c\u8fd9\u4f1a\u524a\u5f31\u7528\u6237\u5bf9\u6a21\u578b\u7684\u4fe1\u5fc3\u3002\u66f4\u7cdf\u7cd5\u7684\u662f\uff0c\u672a\u6765\u7814\u7a76\u7684\u4f5c\u8005\u53ef\u80fd\u4f1a\u88ab\u8bef\u5bfc\uff0c\u8ba4\u4e3a\u4e0d\u5145\u5206\u7684\u6d4b\u8bd5\u662f\u6709\u9053\u7406\u7684\uff0c\u56e0\u4e3a\u5b83\u5df2\u7ecf\u51fa\u73b0\u5728<\/span>\uff08\u53ef\u80fd\u662f\u5df2\u7ecf\u53d1\u8868\u7684\uff09<\/span>\u6570\u7bc7\u7ecf\u8fc7\u540c\u884c\u8bc4\u5ba1\u7684\u6587\u7ae0\u4e2d\u3002\u6b63\u5982\u65b9\u68462\u4e2d\u63d0\u5230\u7684\uff0c\u4f5c\u8005\u3001\u540c\u884c\u8bc4\u5ba1\u548c\u671f\u520a\u7f16\u8f91\u90fd\u6709\u8d23\u4efb\u786e\u4fdd\u907f\u514d\u6570\u636e\u6cc4\u9732\u3002\u6545\u610f\u4fdd\u7559\u8fd9\u4e9b\u7c7b\u578b\u7684\u9519\u8bef\uff0c\u6bd4\u8d77\u5728\u5b9e\u9a8c\u7ed3\u675f\u65f6\u4f2a\u9020\u6570\u636e\uff0c\u786e\u5b9e\u662f\u534a\u65a4\u516b\u4e24\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u6a21\u578b\u7684\u53ef\u89e3\u91ca\u6027<\/strong><\/span>\u3002\u901a\u5e38\u60c5\u51b5\u4e0b\uff0c\u751f\u7269\u5b66\u5bb6\u60f3\u77e5\u9053\u4e3a\u4ec0\u4e48\u7279\u5b9a\u6a21\u578b\u4f1a\u505a\u51fa\u7279\u5b9a\u9884\u6d4b<\/span>\uff08\u5373\u6a21\u578b\u54cd\u5e94\u8f93\u5165\u6570\u636e\u7684\u54ea\u4e9b\u7279\u5f81\u4ee5\u53ca\u5982\u4f55\u54cd\u5e94\uff09<\/span>\u4ee5\u53ca\u4e3a\u4ec0\u4e48\u5b83\u5728\u67d0\u4e9b\u60c5\u51b5\u4e0b\u6709\u6548\uff0c\u800c\u5728\u5176\u5b83\u60c5\u51b5\u4e0b\u65e0\u6548\u3002\u6362\u53e5\u8bdd\u8bf4\uff0c\u751f\u7269\u5b66\u5bb6\u901a\u5e38\u5bf9\u53d1\u73b0\u8d1f\u8d23\u5efa\u6a21\u8f93\u51fa\u7684\u673a\u5236\u548c\u56e0\u7d20\u611f\u5174\u8da3\uff0c\u800c\u4e0d\u4ec5\u4ec5\u5bf9\u5982\u524d\u6240\u8ff0\u7684\u51c6\u786e\u5efa\u6a21\u3002\u89e3\u91ca\u6a21\u578b\u7684\u80fd\u529b\u53d6\u51b3\u4e8e\u4f7f\u7528\u7684\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u548c\u8f93\u5165\u6570\u636e\u3002\u5bf9\u4e8e\u975e\u795e\u7ecf\u7f51\u7edc\u65b9\u6cd5\u6765\u8bf4\uff0c\u89e3\u91ca\u901a\u5e38\u66f4\u5bb9\u6613\uff0c\u56e0\u4e3a\u8fd9\u4e9b\u65b9\u6cd5\u5177\u6709\u66f4\u9002\u5408\u76f4\u63a5\u6709\u610f\u4e49\u89e3\u91ca\u7684\u7279\u5f81\u96c6\uff0c\u5e76\u4e14\u901a\u5e38\u5177\u6709\u8f83\u5c11\u7684\u53ef\u5b66\u4e60\u53c2\u6570\u3002\u4f8b\u5982\uff0c\u5728\u7b80\u5355\u7ebf\u6027\u56de\u5f52\u6a21\u578b\u7684\u60c5\u51b5\u4e0b\uff0c\u5206\u914d\u7ed9\u6bcf\u4e2a\u8f93\u5165\u7279\u5f81\u7684\u53c2\u6570\u76f4\u63a5\u8868\u660e\u8be5\u7279\u5f81\u5982\u4f55\u5f71\u54cd\u9884\u6d4b\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u8bad\u7ec3\u975e\u795e\u7ecf\u7f51\u7edc\u65b9\u6cd5\u7684\u4f4e\u6210\u672c\u610f\u5473\u7740\u8fdb\u884c\u6d88\u878d\u7814\u7a76<\/span>\uff08ablation study\uff09<\/span>\u662f\u53ef\u53d6\u7684\uff0c\u5176\u4e2d\u6d4b\u91cf\u53bb\u9664\u8f93\u5165\u7684\u5b9a\u4e49\u7279\u5f81\u5bf9\u6027\u80fd\u7684\u5f71\u54cd\u3002\u6d88\u878d\u7814\u7a76\u53ef\u4ee5\u63ed\u793a\u54ea\u4e9b\u7279\u5f81\u5bf9\u624b\u5934\u7684\u5efa\u6a21\u4efb\u52a1\u6700\u6709\u7528\uff0c\u5e76\u4e14\u662f\u4e00\u79cd\u53ef\u80fd\u53d1\u73b0\u66f4\u5f3a\u5927\u3001\u66f4\u6709\u6548\u548c\u53ef\u89e3\u91ca\u7684\u6a21\u578b\u7684\u65b9\u6cd5\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u89e3\u91ca\u795e\u7ecf\u7f51\u7edc<\/span>\uff08\u5c24\u5176\u662f\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\uff09<\/span>\u901a\u5e38\u8981\u56f0\u96be\u5f97\u591a\uff0c\u56e0\u4e3a\u6a21\u578b\u4e2d\u7ecf\u5e38\u6709\u5927\u91cf\u7684\u8f93\u5165\u7279\u5f81\u548c\u53c2\u6570\u3002\u8bc6\u522b\u8f93\u5165\u56fe\u50cf\u4e2d\u5bf9\u7279\u5b9a\u5206\u7c7b\u6700\u8d1f\u8d23\u7684\u533a\u57df\u4ecd\u7136\u662f\u6709\u53ef\u80fd\u7684\uff0c\u4f8b\u5982\uff0c\u901a\u8fc7\u6784\u5efa\u663e\u8457\u56fe<\/span>\uff08saliency map\uff09<\/span> [28] \u3002\u5c3d\u7ba1\u663e\u8457\u56fe\u663e\u793a\u4e86\u56fe\u50cf\u7684\u54ea\u4e9b\u533a\u57df\u662f\u91cd\u8981\u7684\uff0c\u4f46\u53ef\u80fd\u66f4\u96be\u4ee5\u786e\u5b9a\u8fd9\u4e9b\u4f4d\u7f6e\u7684\u54ea\u4e9b\u6570\u636e\u5c5e\u6027\u662f\u8d1f\u8d23\u9884\u6d4b\u7684\uff0c\u7279\u522b\u662f\u5f53\u8f93\u5165<\/span>\uff08\u4f8b\u5982\u56fe\u50cf\u548c\u6587\u672c\uff09<\/span>\u4e0d\u5bb9\u6613\u88ab\u4eba\u7c7b\u89e3\u91ca\u65f6\u3002\u5c3d\u7ba1\u5982\u6b64\uff0c\u663e\u8457\u56fe\u548c\u7c7b\u4f3c\u6982\u5ff5\u53ef\u4ee5\u7528\u4f5c\u201c\u5065\u5168\u6027\u68c0\u67e5\u201d\uff0c\u4ee5\u786e\u4fdd\u6a21\u578b\u786e\u5b9e\u5728\u67e5\u770b\u56fe\u50cf\u7684\u76f8\u5173\u90e8\u5206\u3002\u8fd9\u53ef\u4ee5\u5e2e\u52a9\u907f\u514d\u6a21\u578b\u8fdb\u884c\u610f\u5916\u8fde\u63a5\u7684\u60c5\u51b5\uff0c\u4f8b\u5982\uff0c\u6839\u636e\u56fe\u50cf\u89d2\u843d\u7684\u533b\u9662\u6216\u90e8\u95e8\u6807\u7b7e\uff0c\u800c\u975e\u56fe\u50cf\u672c\u8eab\u7684\u533b\u5b66\u5185\u5bb9\uff0c\u5bf9\u533b\u5b66\u56fe\u50cf\u8fdb\u884c\u5206\u7c7b [96]\u3002\u751f\u6210\u5bf9\u6297\u6027\u793a\u4f8b<\/span>\uff08GAN\uff09<\/span>\u6216\u5408\u6210\u8f93\u5165\u53ef\u80fd\u5bfc\u81f4\u795e\u7ecf\u7f51\u7edc\u4ea7\u751f\u81ea\u4fe1\u7684\u9519\u8bef\u9884\u6d4b\uff0c\u4f46\u4e5f\u662f\u53ef\u4ee5\u63d0\u4f9b\u6709\u5173\u54ea\u4e9b\u7279\u5f81\u6700\u5e38\u7528\u4e8e\u9884\u6d4b\u7684\u4fe1\u606f\u7684\u597d\u65b9\u6cd5 [97]\u3002\u4f8b\u5982\uff0cCNN \u7ecf\u5e38\u4f7f\u7528\u7eb9\u7406<\/span>\uff08\u4f8b\u5982\u52a8\u7269\u6bdb\u76ae\u4e2d\u7684\u6761\u7eb9\uff09<\/span>\u6765\u5bf9\u56fe\u50cf\u4e2d\u7684\u5bf9\u8c61\u8fdb\u884c\u5206\u7c7b\uff0c\u800c\u4eba\u7c7b\u4e3b\u8981\u4f7f\u7528\u5f62\u72b6 [51]\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\n
\n
\n

<\/section>\n

<\/section>\n<\/section>\n
\n
\n

\u65b9\u68462 | \u8bc4\u4f30\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u7684\u6587\u7ae0<\/span><\/strong><\/span><\/p>\n


<\/span><\/p>\n

\u4ee5\u4e0b\u662f\u5728\u9605\u8bfb\u6216\u5ba1\u9605\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u5904\u7406\u751f\u7269\u6570\u636e\u7684\u6587\u7ae0\u65f6\u9700\u8981\u8003\u8651\u7684\u4e00\u4e9b\u95ee\u9898\u3002\u5373\u4f7f\u4e0d\u4e00\u5b9a\u5f97\u5230\u6709\u6548\u56de\u7b54\uff0c\u8bb0\u4f4f\u8fd9\u4e9b\u95ee\u9898\u4e5f\u662f\u6709\u7528\u7684\uff0c\u5e76\u4e14\u8fd9\u4e9b\u95ee\u9898\u53ef\u4ee5\u7528\u4f5c\u4e0e\u673a\u5668\u5b66\u4e60\u80cc\u666f\u5408\u4f5c\u8005\u8fdb\u884c\u8ba8\u8bba\u7684\u57fa\u7840\u3002\u6709\u60ca\u4eba\u6570\u91cf\u7684\u6587\u7ae0\u5e76\u4e0d\u80fd\u8fbe\u5230\u8fd9\u4e9b\u6807\u51c6[148]\u3002<\/span><\/p>\n

 <\/span><\/p>\n

\u662f\u5426\u5145\u5206\u63cf\u8ff0\u4e86\u6570\u636e\u96c6\uff1f<\/span><\/strong><\/p>\n

\u7814\u7a76\u4eba\u5458\u5e94\u63d0\u4f9b\u6784\u5efa\u6570\u636e\u96c6\u7684\u5b8c\u6574\u6b65\u9aa4\uff0c\u6700\u597d\u4f7f\u7528\u5728\u7f51\u9875\u4e2d\u6c38\u4e45\u4fdd\u7559\u7684\u6570\u636e\u96c6\u6216\u6458\u8981\u6570\u636e\u3002\u6839\u636e\u6211\u4eec\u7684\u7ecf\u9a8c\uff0c\u5bf9\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u8fdb\u884c\u5168\u9762\u63cf\u8ff0\uff0c\u4f46\u5bf9\u6570\u636e\u7684\u63cf\u8ff0\u5374\u542b\u7cca\u5176\u8f9e\uff0c\u662f\u5371\u9669\u4fe1\u53f7\uff08red flag\uff09\u3002\u5982\u679c\u4f7f\u7528\u7684\u6570\u636e\u96c6\u662f\u6807\u51c6\u6570\u636e\u96c6\uff0c\u6216\u6765\u81ea\u53e6\u4e00\u9879\u7814\u7a76\uff0c\u5219\u5e94\u5728\u6587\u7ae0\u4e2d\u660e\u786e\u8bf4\u660e\u3002<\/span><\/p>\n

 <\/span><\/p>\n

\u6d4b\u8bd5\u96c6\u662f\u5426\u786e\u51ff\uff1f<\/span><\/strong><\/p>\n

\u6839\u636e\u751f\u7269\u5e94\u7528\u6311\u6218\u90e8\u5206\u4e2d\u7684\u8ba8\u8bba\uff0c\u68c0\u67e5\u6d4b\u8bd5\u96c6\u662f\u5426\u8db3\u4ee5\u5bf9\u6240\u8c03\u67e5\u7684\u5c5e\u6027\u8fdb\u884c\u57fa\u51c6\u6d4b\u8bd5\u3002\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u4e4b\u95f4\u4e0d\u5e94\u8be5\u6709\u6570\u636e\u6cc4\u6f0f\uff0c\u6d4b\u8bd5\u96c6\u5e94\u8be5\u8db3\u591f\u5927\u4ee5\u63d0\u4f9b\u53ef\u9760\u7684\u7ed3\u679c\uff0c\u5e76\u4e14\u6d4b\u8bd5\u96c6\u5e94\u8be5\u8986\u76d6\u8be5\u5de5\u5177\u7684\u7528\u6237\u53ef\u80fd\u7684\u4f7f\u7528\u8303\u56f4\u3002\u7814\u7a76\u4eba\u5458\u540c\u6837\u5e94\u8be5\u8be6\u7ec6\u8ba8\u8bba\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u7684\u7ec4\u6210\u548c\u5927\u5c0f\u3002\u4f5c\u8005\u6709\u8d23\u4efb\u786e\u4fdd\u5df2\u91c7\u53d6\u6240\u6709\u6b65\u9aa4\u4ee5\u907f\u514d\u6570\u636e\u6cc4\u9732\uff0c\u5e76\u4e14\u5e94\u5728\u6587\u7ae0\u4e2d\u63cf\u8ff0\u8fd9\u4e9b\u6b65\u9aa4\u53ca\u5176\u80cc\u540e\u7684\u57fa\u672c\u539f\u7406\u3002\u5b66\u672f\u671f\u520a\u7f16\u8f91\u548c\u5ba1\u7a3f\u4eba\u8fd8\u5e94\u8be5\u786e\u4fdd\u8fd9\u4e9b\u4efb\u52a1\u5df2\u7ecf\u6309\u7167\u826f\u597d\u7684\u6807\u51c6\u6267\u884c\uff0c\u800c\u975e\u4ec5\u9760\u8bba\u6587\u4f5c\u8005\u81ea\u89c9\u3002<\/span><\/p>\n

 <\/span><\/p>\n

\u6a21\u578b\u9009\u62e9\u662f\u5426\u5408\u7406\uff1f<\/span><\/strong><\/p>\n

\u7814\u7a76\u4eba\u5458\u5e94\u7ed9\u51fa\u9009\u62e9\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u7684\u7406\u7531\u3002\u4f7f\u7528\u795e\u7ecf\u7f51\u7edc\u7684\u7406\u7531\u662f\u56e0\u4e3a\u5b83\u4eec\u9002\u7528\u4e8e\u5f53\u524d\u6570\u636e\u548c\u95ee\u9898\uff0c\u800c\u975e\u4ec5\u4ec5\u662f\u56e0\u4e3a\u5176\u4ed6\u4eba\u90fd\u5728\u4f7f\u7528\u3002\u5b66\u672f\u754c\u5e94\u9f13\u52b1\u8ba8\u8bba\u5df2\u5c1d\u8bd5\u4f46\u65e0\u6548\u7684\u6a21\u578b\uff0c\u56e0\u4e3a\u5b83\u53ef\u4ee5\u5e2e\u52a9\u5176\u4ed6\u7814\u7a76\u4eba\u5458\uff1b\u5f88\u591a\u65f6\u5019\uff0c\u4e00\u4e2a\u590d\u6742\u7684\u6a21\u578b\u6ca1\u6709\u5bf9\u6700\u7ec8\u5f97\u5230\u8be5\u6a21\u578b\u6240\u9700\u7684\u4e0d\u53ef\u907f\u514d\u7684\u53cd\u590d\u8bd5\u9a8c\u8fdb\u884c\u4efb\u4f55\u8ba8\u8bba\uff0c\u4f46\u4f9d\u7136\u5f97\u4ee5\u53d1\u8868\u3002<\/span><\/p>\n

 <\/span><\/p>\n

\u8be5\u65b9\u6cd5\u662f\u5426\u4e0e\u5176\u4ed6\u65b9\u6cd5\u8fdb\u884c\u4e86\u6bd4\u8f83\uff1f<\/span><\/strong><\/p>\n

\u5e94\u8be5\u5c06\u4e00\u79cd\u65b0\u65b9\u6cd5\u4e0e\u8868\u73b0\u51fa\u826f\u597d\u6027\u80fd\u5e76\u88ab\u5e7f\u6cdb\u4f7f\u7528\u7684\u73b0\u6709\u65b9\u6cd5\u8fdb\u884c\u6bd4\u8f83\u3002\u7406\u60f3\u60c5\u51b5\u4e0b\uff0c\u5e94\u6bd4\u8f83\u4f7f\u7528\u5404\u79cd\u6a21\u578b\u7c7b\u578b\u7684\u65b9\u6cd5\uff0c\u8fd9\u6837\u6709\u52a9\u4e8e\u89e3\u91ca\u7ed3\u679c\u3002\u4ee4\u4eba\u60ca\u8bb6\u7684\u662f\uff0c\u8bb8\u591a\u590d\u6742\u6a21\u578b\u5728\u6027\u80fd\u4e0a\u5176\u5b9e\u901a\u8fc7\u7b80\u5355\u7684\u56de\u5f52\u65b9\u6cd5\u5c31\u53ef\u4ee5\u5339\u914d\u3002<\/span><\/p>\n

 <\/span><\/p>\n

\u7ed3\u679c\u597d\u5f97\u4ee4\u4eba\u96be\u4ee5\u7f6e\u4fe1\u5417\uff1f<\/span><\/strong><\/p>\n

\u8d85\u8fc7 99% \u51c6\u786e\u7387\u7684\u58f0\u660e\uff0c\u5728\u751f\u7269\u5b66\u7684\u673a\u5668\u5b66\u4e60\u6587\u7ae0\u4e2d\u5e76\u4e0d\u5c11\u89c1\u3002\u901a\u5e38\uff0c\u8fd9\u662f\u6d4b\u8bd5\u51fa\u73b0\u95ee\u9898\u7684\u8ff9\u8c61\uff0c\u800c\u4e0d\u662f\u60ca\u4eba\u7684\u7a81\u7834\u3002\u4f5c\u8005\u548c\u5ba1\u7a3f\u4eba\u90fd\u5e94\u8be5\u6ce8\u610f\u8fd9\u4e00\u70b9\u3002<\/span><\/p>\n

 <\/span><\/p>\n

\u65b9\u6cd5\u53ef\u7528\u5417\uff1f<\/span><\/strong><\/p>\n

\u81f3\u5c11\uff0c\u60f3\u8981\u4f7f\u7528\u6587\u7ae0\u4e2d\u8bad\u7ec3\u597d\u6a21\u578b\u7684\u79d1\u7814\u4eba\u5458\uff0c\u5e94\u8be5\u80fd\u591f\u57fa\u4e8e\u7f51\u9875\u670d\u52a1\u6216\u4ee3\u7801\u6587\u4ef6\u8fd0\u884c\u4e00\u6b21\u9884\u6d4b\u3002\u7406\u60f3\u60c5\u51b5\u4e0b\uff0c\u81f3\u5c11\u5728\u6c38\u4e45 URL \u548c\u901a\u7528\u8bb8\u53ef\u4e0b\uff0c\u5e94\u8be5\u63d0\u4f9b\u6e90\u4ee3\u7801\u548c\u7ecf\u8fc7\u8bad\u7ec3\u7684\u6a21\u578b[149,150]\u3002\u4f7f\u8bad\u7ec3\u4ee3\u7801\u53ef\u7528\uff0c\u4e5f\u662f\u7406\u60f3\u60c5\u51b5\u4e0b\u8be5\u6709\u7684\uff0c\u56e0\u4e3a\u8fd9\u8fdb\u4e00\u6b65\u63d0\u9ad8\u4e86\u6587\u7ae0\u7684\u53ef\u91cd\u590d\u6027\u5e76\u5141\u8bb8\u5176\u4ed6\u7814\u7a76\u4eba\u5458\u4ee5\u8be5\u65b9\u6cd5\u4e3a\u57fa\u7840\uff0c\u800c\u65e0\u9700\u4ece\u5934\u5f00\u59cb\u3002\u671f\u520a\u5e94\u8be5\u5728\u8fd9\u91cc\u627f\u62c5\u4e00\u4e9b\u8d23\u4efb\uff0c\u4ee5\u786e\u4fdd\u8fd9\u6210\u4e3a\u5e38\u6001\u3002<\/span><\/p>\n<\/section>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n


<\/p>\n

\u4fdd\u62a4\u9690\u79c1\u7684\u673a\u5668\u5b66\u4e60<\/span><\/strong>\u3002\u4e00\u4e9b\u751f\u7269\u6570\u636e\uff0c\u5c24\u5176\u662f\u4eba\u7c7b\u57fa\u56e0\u7ec4\u6570\u636e\u548c\u5546\u4e1a\u654f\u611f\u7684\u836f\u7269\u6570\u636e\uff0c\u5177\u6709\u6570\u636e\u9690\u79c1\u65b9\u9762\u7684\u5f71\u54cd\u3002\u5728\u6570\u636e\u9690\u79c1\u7684\u80cc\u666f\u4e0b\uff0c\u7814\u7a76\u4eba\u5458\u5df2\u7ecf\u505a\u51fa\u4e86\u8bb8\u591a\u52aa\u529b\uff0c\u6765\u5b9e\u73b0\u6570\u636e\u5171\u4eab\u548c\u673a\u5668\u5b66\u4e60\u6a21\u578b\u7684\u5206\u5e03\u5f0f\u8bad\u7ec3\u3002\u4f8b\u5982\uff0c\u73b0\u4ee3\u5bc6\u7801\u6280\u672f\u5141\u8bb8\u5728\u6570\u636e\u548c\u7ed3\u679c\u53ef\u8bc1\u660e\u662f\u5b89\u5168\u7684\u60c5\u51b5\u4e0b\uff0c\u8bad\u7ec3\u836f\u7269-\u9776\u6807\u76f8\u4e92\u4f5c\u7528\u6a21\u578b[98]\u3002\u4e0e\u4e34\u5e8a\u8bd5\u9a8c\u4e2d\u7684\u771f\u5b9e\u53c2\u4e0e\u8005\u975e\u5e38\u76f8\u4f3c\u7684\u6a21\u62df\u5408\u6210\u53c2\u4e0e\u8005\uff0c\u53ef\u4ee5\u5728\u4e0d\u6cc4\u9732\u8bc6\u522b\u6570\u636e\u7684\u524d\u63d0\u4e0b\uff0c\u5f97\u51fa\u5bf9\u771f\u5b9e\u53c2\u4e0e\u8005\u51c6\u786e\u7684\u7ed3\u679c[99]\u3002\u7814\u7a76\u4eba\u5458\u5df2\u7ecf\u5f00\u53d1\u4e86\u4e00\u4e9b\u7b97\u6cd5\uff0c\u53ef\u4ee5\u4f7f\u7528\u5b58\u50a8\u5728\u4e0d\u540c\u4f4d\u7f6e\u7684\u6570\u636e\uff0c\u8fdb\u884c\u6709\u6548\u7684\u8054\u5408\u6a21\u578b\u8bad\u7ec3 [100]\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u8de8\u5b66\u79d1\u5408\u4f5c\u7684\u5fc5\u8981\u6027<\/span><\/strong>\u3002\u9664\u975e\u4f7f\u7528\u516c\u5f00\u53ef\u7528\u7684\u6570\u636e\uff0c\u5426\u5219\u4e00\u4e2a\u7814\u7a76\u7ec4\u5f88\u5c11\u4f1a\u540c\u65f6\u62e5\u6709\u4e13\u4e1a\u77e5\u8bc6\u548c\u8d44\u6e90\u2014\u2014\u65e2\u53ef\u4ee5\u4e3a\u673a\u5668\u5b66\u4e60\u7814\u7a76\u6536\u96c6\u6570\u636e\uff0c\u53c8\u53ef\u4ee5\u6709\u6548\u5730\u5e94\u7528\u6700\u5408\u9002\u7684\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u3002\u5b9e\u9a8c\u751f\u7269\u5b66\u5bb6\u4e0e\u8ba1\u7b97\u673a\u79d1\u5b66\u5bb6\u5408\u4f5c\u5341\u5206\u5e38\u89c1\uff0c\u800c\u4e14\u901a\u5e38\u4f1a\u53d6\u5f97\u5f88\u597d\u7684\u7ed3\u679c\u3002\u7136\u800c\uff0c\u5728\u8fd9\u79cd\u5408\u4f5c\u4e2d\uff0c\u91cd\u8981\u7684\u662f\u6bcf\u4e00\u65b9\u90fd\u4e86\u89e3\u5bf9\u65b9\u7684\u4e00\u4e9b\u5de5\u4f5c\u77e5\u8bc6\u3002\u7279\u522b\u662f\u8ba1\u7b97\u673a\u79d1\u5b66\u5bb6\u5e94\u8be5\u52aa\u529b\u4e86\u89e3\u6570\u636e\u2014\u2014\u4f8b\u5982\u9884\u671f\u7684\u566a\u58f0\u7a0b\u5ea6\u548c\u53ef\u91cd\u590d\u6027\u2014\u2014\u540c\u65f6\u751f\u7269\u5b66\u5bb6\u5e94\u8be5\u4e86\u89e3\u6240\u4f7f\u7528\u7684\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u7684\u5c40\u9650\u6027\u3002\u5efa\u7acb\u8fd9\u6837\u7684\u7406\u89e3\u9700\u8981\u65f6\u95f4\u548c\u7cbe\u529b\uff0c\u4f46\u5bf9\u4e8e\u9632\u6b62\u4e0d\u826f\u6a21\u578b\u548c\u8bef\u5bfc\u6027\u7ed3\u679c\u7684\u65e0\u610f\u4f20\u64ad\u5f88\u91cd\u8981\u3002<\/span><\/section>\n
 <\/span><\/section>\n
       <\/span><\/section>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

\u672a\u6765\u53d1\u5c55\u65b9\u5411<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n


<\/p>\n

\u5728\u53ef\u9884\u89c1\u7684\u672a\u6765\uff0c\u673a\u5668\u5b66\u4e60\u5728\u751f\u7269\u5b66\u7814\u7a76\u4e2d\u7684\u4f7f\u7528\u8d8a\u6765\u8d8a\u591a\u7684\u8d8b\u52bf\uff0c\u770b\u8d77\u6765\u4e5f\u5c06\u7ee7\u7eed\u4e0b\u53bb\u3002\u65b9\u6cd5\u8bba\u3001\u8f6f\u4ef6\u548c\u786c\u4ef6\u7684\u91cd\u8981\u8fdb\u6b65\u4f7f\u8fd9\u79cd\u5438\u6536\u589e\u52a0\u6210\u4e3a\u53ef\u80fd\uff0c\u6240\u6709\u65b9\u5411\u90fd\u5728\u4e0d\u65ad\u53d1\u5c55\u3002\u8bb8\u591a\u5927\u578b\u79d1\u6280\u516c\u53f8\u6b63\u5728\u5229\u7528\u4ed6\u4eec\u7684\u6280\u672f\u4e13\u957f\u548c\u5927\u91cf\u8d44\u6e90\u6765\u534f\u52a9\u5b66\u672f\u7814\u7a76\u4eba\u5458\uff0c\u751a\u81f3\u901a\u8fc7\u521b\u65b0\u7684\u673a\u5668\u5b66\u4e60\u7b56\u7565\u8fdb\u884c\u751f\u7269\u5b66\u7814\u7a76\u3002\u7136\u800c\uff0c\u8fc4\u4eca\u4e3a\u6b62\uff0c\u5927\u591a\u6570\u6210\u529f\u6765\u81ea\u5c06\u5176\u4ed6\u9886\u57df\u5f00\u53d1\u7684\u7b97\u6cd5\u76f4\u63a5\u5e94\u7528\u4e8e\u751f\u7269\u6570\u636e\u3002\u4f8b\u5982\uff0cCNN \u548c RNN \u5206\u522b\u66f4\u591a\u5e94\u7528\u4e8e\u56fe\u50cf\u5206\u6790<\/span>\uff08\u7528\u4e8e\u4eba\u8138\u8bc6\u522b\u6216\u81ea\u52a8\u9a7e\u9a76\u6c7d\u8f66\uff09<\/span>\u548c\u81ea\u7136\u8bed\u8a00\u5904\u7406\u7b49\u5e94\u7528\u3002\u751f\u7269\u5b66\u673a\u5668\u5b66\u4e60\u6700\u4ee4\u4eba\u5174\u594b\u7684\u524d\u666f\u4e4b\u4e00\uff0c\u662f\u4e13\u95e8\u9488\u5bf9\u751f\u7269\u5b66\u6570\u636e\u548c\u751f\u7269\u5b66\u95ee\u9898\u91cf\u8eab\u5b9a\u5236\u7684\u7b97\u6cd5<\/strong> [101,102] \u3002\u5982\u679c\u53ef\u4ee5\u5229\u7528\u751f\u7269\u7cfb\u7edf\u7684\u5df2\u77e5\u7ed3\u6784\u5e76\u4f7f\u7528\u795e\u7ecf\u7f51\u7edc\u6765\u5b66\u4e60\u672a\u77e5\u90e8\u5206\uff0c\u5c31\u53ef\u4ee5\u7528\u66f4\u6613\u4e8e\u89e3\u91ca\u4e14\u5bf9\u65b0\u6570\u636e\u66f4\u7a33\u5065\u7684\u66f4\u7b80\u5355\u7684\u6a21\u578b\uff0c\u6765\u53d6\u4ee3\u53c2\u6570\u5316\u8d8a\u6765\u8d8a\u4e25\u91cd\u7684\u6a21\u578b [103]\u3002\u5e94\u7528\u5305\u62ec\u751f\u7269\u53cd\u5e94\u7cfb\u7edf\u548c\u836f\u4ee3\u52a8\u529b\u5b66\uff0c\u5728\u8fd9\u4e9b\u5e94\u7528\u4e2d\u53ef\u4ee5\u4f7f\u7528\u5df2\u77e5\u5fae\u5206\u65b9\u7a0b\u7cfb\u7edf\u3002\u8fd9\u4e5f\u5c06\u6709\u52a9\u4e8e\u4ece\u9884\u6d4b\u673a\u5668\u5b66\u4e60\u8f6c\u5411\u53ef\u4ee5\u521b\u5efa\u65b0\u5b9e\u4f53\u7684\u751f\u6210\u6a21\u578b\uff0c\u4f8b\u5982\u8bbe\u8ba1\u5177\u6709\u65b0\u7ed3\u6784\u548c\u529f\u80fd\u7684\u86cb\u767d\u8d28 [104,105]\u3002<\/span>
<\/section>\n
       <\/span><\/section>\n
\u968f\u7740\u6709\u7528\u67b6\u6784\u548c\u8f93\u5165\u6570\u636e\u7c7b\u578b\u79cd\u7c7b\u7684\u589e\u52a0\uff0c\u53ef\u5fae\u7f16\u7a0b\u7684\u8303\u5f0f\u6b63\u5728\u4ece\u6df1\u5ea6\u5b66\u4e60\u9886\u57df\u51fa\u73b0 [106]\u3002\u53ef\u5fae\u7f16\u7a0b\u662f\u4f7f\u7528\u81ea\u52a8\u5fae\u5206<\/span>\uff08\u8bad\u7ec3\u795e\u7ecf\u7f51\u7edc\u7684\u6838\u5fc3\u6982\u5ff5\uff09<\/span>\u6765\u8ba1\u7b97\u68af\u5ea6\u548c\u6539\u8fdb\u4efb\u4f55\u6240\u9700\u7b97\u6cd5\u4e2d\u7684\u53c2\u6570\u3002\u8fd9\u663e\u793a\u4e86\u86cb\u767d\u8d28\u7ed3\u6784\u9884\u6d4b\u4e2d\u751f\u7269\u7cfb\u7edf\u7269\u7406\u6a21\u578b\u7684\u524d\u666f [63,107]\uff0c\u4ee5\u53ca\u5b66\u4e60\u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df\u7684\u529b\u573a\u53c2\u6570 [108,109]\u3002\u53ef\u5fae\u8f6f\u4ef6\u5305<\/span>\uff08\u5982 JAX [110]\uff09<\/span>\u548c\u9488\u5bf9\u7279\u5b9a\u751f\u7269\u5b66\u9886\u57df\u7684\u8f6f\u4ef6\u5305<\/span>\uff08\u5982 Selene [111]\u3001Janggu [112] \u548c JAX MD [113]\uff09<\/span>\u7684\u5f00\u53d1\u5c06\u6709\u52a9\u4e8e\u6b64\u7c7b\u65b9\u6cd5\u7684\u5f00\u53d1\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u8fdb\u884c\u751f\u7269\u6570\u636e\u5206\u6790\u7684\u8fdb\u5c55\u4e5f\u5f97\u76ca\u4e8e\u5728\u516c\u5f00\u53ef\u7528\u7684\u5b58\u50a8\u5e93\u4e2d\u5b58\u653e\u8bad\u7ec3\u597d\u7684\u6a21\u578b\u3002\u8fd9\u6837\uff0c\u7814\u7a76\u7c7b\u4f3c\u95ee\u9898\u7684\u7814\u7a76\u4eba\u5458\u53ef\u4ee5\u4f7f\u7528\u8fd9\u4e9b\u6a21\u578b\u800c\u65e0\u9700\u8bad\u7ec3\uff0c\u5e76\u4e14\u53ef\u4ee5\u4f7f\u7528\u5404\u79cd\u4e0d\u540c\u7684\u6a21\u578b\uff0c\u53ea\u9700\u6781\u5c11\u7684\u5de5\u4f5c\u5373\u53ef\u5728\u5b83\u4eec\u4e4b\u95f4\u5207\u6362[114]\u3002\u8be5\u9886\u57df\u8fd8\u51fa\u73b0\u4e86\u4e00\u79cd\u6269\u5c55\u2014\u2014\u81ea\u52a8\u5316\u673a\u5668\u5b66\u4e60\u7ba1\u9053\uff0c\u65e0\u9700\u7528\u6237\u8f93\u5165\u5373\u53ef\u8bad\u7ec3\u548c\u8c03\u6574\u5404\u79cd\u6a21\u578b\u5e76\u8fd4\u56de\u6027\u80fd\u6700\u4f73\u7684\u6a21\u578b<\/strong>\u3002\u8fd9\u4e9b\u7ba1\u9053\u53ef\u4ee5\u5e2e\u52a9\u975e\u4e13\u5bb6\u8bad\u7ec3\u6a21\u578b[115]\u3002\u7136\u800c\uff0c\u8fd9\u4e9b\u8d44\u6e90\u4e0d\u80fd\u66ff\u4ee3\u5bf9\u6240\u7528\u65b9\u6cd5\u7684\u900f\u5f7b\u7406\u89e3\uff0c\u5bf9\u4e8e\u9009\u62e9\u9002\u5f53\u7684\u5f52\u7eb3\u504f\u5dee\u548c\u89e3\u91ca\u6a21\u578b\u7684\u9884\u6d4b\u5f88\u91cd\u8981\u3002\u81f3\u4e8e\u81ea\u52a8\u5316\u673a\u5668\u5b66\u4e60\u662f\u5426\u5df2\u7ecf\u53ef\u9760\u548c\u7075\u6d3b\u5230\u5141\u8bb8\u5b9e\u9a8c\u4eba\u5458\u72ec\u7acb\u5730\u5e38\u89c4\u4f7f\u7528\u590d\u6742\u7684\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\uff0c\u6216\u8005\u8bf4\uff0c\u673a\u5668\u5b66\u4e60\u4e13\u4e1a\u77e5\u8bc6\u662f\u5426\u4ecd\u7136\u662f\u5fc5\u9700\u7684\uff0c\u8fd8\u6709\u5f85\u672a\u6765\u89c2\u5bdf\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u5982\u524d\u6240\u8ff0\uff0c\u6a21\u578b\u7684\u4e25\u683c\u9a8c\u8bc1\u548c\u4e0d\u540c\u6a21\u578b\u7684\u6bd4\u8f83\u5177\u6709\u6311\u6218\u6027\uff0c\u4f46\u5bf9\u4e8e\u786e\u5b9a\u6027\u80fd\u6700\u4f73\u7684\u6a21\u578b\u548c\u4e3a\u672a\u6765\u7684\u7814\u7a76\u65b9\u5411\u63d0\u4f9b\u4fe1\u606f\u4ecd\u7136\u662f\u5fc5\u8981\u7684\u3002\u4e3a\u4e86\u4f7f\u8be5\u9886\u57df\u53d6\u5f97\u8fdb\u5c55\uff0c\u6709\u5fc5\u8981\u5f00\u53d1\u57fa\u51c6\u6570\u636e\u96c6\u548c\u9a8c\u8bc1\u4efb\u52a1 [116]\uff0c\u4f8b\u5982 Protein Net [117]\u3001ATOM3D [118] \u548c TAPE [119]\uff0c\u5e76\u4f7f\u5176\u5f97\u5230\u5e7f\u6cdb\u5e94\u7528\u3002\u5f53\u7136\uff0c\u53ef\u80fd\u4f1a\u53d1\u751f\u9488\u5bf9\u7279\u5b9a\u57fa\u51c6\u7684\u8fc7\u5ea6\u4f18\u5316\uff0c\u6240\u4ee5\u91cd\u8981\u7684\u662f\u9700\u8981\u7814\u7a76\u4eba\u5458\u62b5\u5236\u8fd9\u79cd\u80fd\u4f7f\u4ed6\u4eec\u7684\u7ed3\u679c\u770b\u8d77\u6765\u66f4\u597d\u7684\u8bf1\u60d1\u3002\u8bf8\u5982 CASP [120] \u548c\u529f\u80fd\u6ce8\u91ca\u7684\u6279\u5224\u6027\u8bc4\u4f30 [121] \u7b49\u76f2\u8bc4\u4f30\uff0c\u5c06\u7ee7\u7eed\u5728\u8bc4\u4f30\u54ea\u4e9b\u6a21\u578b\u8868\u73b0\u6700\u4f73\u65b9\u9762\uff0c\u53d1\u6325\u91cd\u8981\u4f5c\u7528\u3002<\/span><\/section>\n
       <\/span><\/section>\n
\u603b\u4f53\u800c\u8a00\uff0c\u751f\u7269\u6570\u636e\u7684\u591a\u6837\u6027\u4f7f\u5f97\u673a\u5668\u5b66\u4e60\u5f88\u96be\u4e3a\u5176\u751f\u7269\u5b66\u4e2d\u7684\u5e94\u7528\u63d0\u4f9b\u901a\u7528\u6307\u5357\u3002\u56e0\u6b64\uff0c\u672c\u6587\u7684\u76ee\u6807\u662f\u4e3a\u751f\u7269\u5b66\u5bb6\u6982\u8ff0\u53ef\u7528\u7684\u4e0d\u540c\u65b9\u6cd5\uff0c\u5e76\u4e3a\u4ed6\u4eec\u63d0\u4f9b\u4e00\u4e9b\u5173\u4e8e\u5982\u4f55\u5728\u4ed6\u4eec\u7684\u6570\u636e\u4e0a\u8fdb\u884c\u6709\u6548\u673a\u5668\u5b66\u4e60\u7684\u9884\u5224\u3002\u5f53\u7136\uff0c\u673a\u5668\u5b66\u4e60\u5e76\u4e0d\u9002\u5408\u6240\u6709\u95ee\u9898\uff0c\u77e5\u9053\u5b83\u4ec0\u4e48\u65f6\u5019\u53d8\u4e3a\u52a3\u52bf\u540c\u6837\u91cd\u8981\uff1a\u5728\u6ca1\u6709\u8db3\u591f\u7684\u6570\u636e\u65f6\u3001\u5728\u9700\u8981\u7406\u89e3\u800c\u975e\u9884\u6d4b\u65f6\uff0c\u6216\u8005\u5728\u4e0d\u6e05\u695a\u5982\u4f55\u8bc4\u4f30\u6027\u80fd\u65f6\u3002\u673a\u5668\u5b66\u4e60\u5728\u751f\u7269\u5b66\u4e2d\u4f55\u65f6\u80fd\u53d1\u6325\u6700\u5927\u6548\u80fd\u7684\u754c\u9650\uff0c\u4ecd\u6709\u5f85\u63a2\u7d22\uff0c\u5e76\u5c06\u6839\u636e\u53ef\u7528\u5b9e\u9a8c\u6570\u636e\u7684\u6027\u8d28\u548c\u6570\u91cf\u7ee7\u7eed\u53d8\u5316\u3002\u4f46\u4e0d\u53ef\u5426\u8ba4\u7684\u662f\uff0c\u673a\u5668\u5b66\u4e60\u5df2\u7ecf\u5bf9\u751f\u7269\u5b66\u4ea7\u751f\u4e86\u5de8\u5927\u5f71\u54cd\uff0c\u5e76\u5c06\u7ee7\u7eed\u5982\u6b64\u3002<\/span><\/section>\n
 <\/span><\/section>\n
 <\/span><\/section>\n
\n
\n
\n

\u53c2\u8003\u6587\u732e<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n

\n
\n
\n
\n

<\/section>\n
1. Ching, T. et al. Opportunities and obstacles for deep learning in biology and medicine. J. R. Soc. Interface 15, 20170387 (2018).<\/span><\/section>\n
2.Mitchell, T. M. Machine Learning (McGraw Hill, 1997).<\/span><\/section>\n
3.Goodfellow, I., Bengio Y. & Courville, A. Deep Learning (MIT Press, 2016).<\/span><\/section>\n
4. Libbrecht, M. W. & Noble, W. S. Machine learning applications in genetics and genomics. Nat. Rev. Genet.16, 321\u2013332 (2015).<\/span><\/section>\n
5. Zou, J. et al. A primer on deep learning in genomics.Nat. Genet. 51, 12\u201318 (2019).<\/span><\/section>\n
6.Myszczynska, M. A. et al. Applications of machine learning to diagnosis and treatment of neurodegenerative diseases. Nat. Rev. Neurol. 16,440\u2013456 (2020).<\/span><\/section>\n
7.Yang, K. K., Wu, Z. & Arnold, F. H. Machine- learning-guided directed evolution for protein engineering. <\/span>Nat. Methods 16, 687\u2013694 (2019).<\/span><\/section>\n
8.Tarca, A. L., Carey, V. J., Chen, X.-W., Romero, R.& Dr\u0103ghici, S. Machine learning and its applications <\/span>to biology. PLoS Comput. Biol. 3, e116 (2007).This is an introduction to machine learning concepts and applications in biology with a focus on traditional machine learning methods.<\/span><\/section>\n
9.Silva, J. C. F., Teixeira, R. M., Silva, F. F.,Brommonschenkel, S. H. & Fontes, E. P. B. Machine learning approaches and their current application in plant molecular biology: a systematic review.Plant. Sci. 284, 37\u201347 (2019).<\/span><\/section>\n
10. Kandoi, G., Acencio, M. L. & Lemke, N. Prediction of druggable proteins using machine learning and systems biology: a mini- review. Front. Physiol. 6, 366 (2015).<\/span><\/section>\n
11. Marblestone, A. H., Wayne, G. & Kording, K. P. Toward an integration of deep learning and neuroscience.Front. Comput. Neurosci. 10, 94 (2016).<\/span><\/section>\n
12. Jim\u00e9nez- Luna, J., Grisoni, F. & Schneider, G.Drug discovery with explainable artificial intelligence. <\/span>Nat. Mach. Intell. 2, 573\u2013584 (2020).<\/span><\/section>\n
13. Buchan, D. W. A. & Jones, D. T. The PSIPRED Protein Analysis Workbench: 20 years on. Nucleic Acids Res.47, W402\u2013W407 (2019).<\/span><\/section>\n
14. Kelley, D. R., Snoek, J. & Rinn, J. L. Basset: learning the regulatory code of the accessible genome with deep convolutional neural networks. Genome Res. 26,990\u2013999 (2016).<\/span><\/section>\n
15. Altman, N. & Krzywinski, M. Clustering. Nat. Methods 14, 545\u2013546 (2017).<\/span><\/section>\n
16. Hopf, T. A. et al. Mutation effects predicted from sequence co- variation. Nat. Biotechnol. 35, 128\u2013135(2017).<\/span><\/section>\n
17. Zhang, Z. et al. Predicting folding free energy changes upon single point mutations. Bioinformatics 28,664\u2013671 (2012).<\/span><\/section>\n
18. Pedregosa, F. et al. Scikit- learn: machine learning in python. J. Mach. Learn. Res. 12, 2825\u20132830 (2011).<\/span><\/section>\n
19. Kuhn, M. Building predictive models in r using the caret package. J. Stat. Softw. 28, 1\u201326 (2008).<\/span><\/section>\n
20. Blaom, A. D. et al. MLJ: a Julia package for composable machine learning. J. Open Source Softw.<\/span>5, 2704 (2020).<\/span><\/section>\n
21. Jones, D. T. Setting the standards for machine learning in biology. Nat. Rev. Mol. Cell Biol. 20, 659\u2013660(2019).<\/span><\/section>\n
22. Alipanahi, B., Delong, A., Weirauch, M. T. & Frey, B. J. Predicting the sequence specificities of DNA- and RNA- binding proteins by deep learning.Nat. Biotechnol. 33, 831\u2013838 (2015).<\/span><\/section>\n
23. Senior, A. W. et al. Improved protein structure prediction using potentials from deep learning.Nature 577, 706\u2013710 (2020).<\/span><\/section>\n
Technology company DeepMind entered the CASP13 assessment in protein structure prediction and its method using deep learning was the most accurate of the methods entered.<\/span><\/section>\n
24. Esteva, A. et al. Dermatologist- level classification of skin cancer with deep neural networks. Nature 542,115\u2013118 (2017).<\/span><\/section>\n
25. Tegunov, D. & Cramer, P. Real- time cryo- electron microscopy data preprocessing with Warp.Nat. Methods 16, 1146\u20131152 (2019).<\/span><\/section>\n
26. LeCun, Y., Bengio, Y. & Hinton, G. Deep learning. Nature 521, 436\u2013444 (2015). <\/span>This is a review of deep learning by some of the major figures in the deep learning revolution.<\/span><\/section>\n
27. Hastie T., Tibshirani R., Friedman J. The elements of statistical learning: data mining, inference, and prediction. 2nd Edn. (Springer Science & Business Media; 2009).<\/span><\/section>\n
28. Adebayo, J. et al. <\/span>Sanity checks for saliency maps. NeurIPS <\/span>https:\/\/arxiv.org\/abs\/1810.03292 (2018).<\/span><\/span><\/section>\n
29. Gal, Y. & Ghahramani, Z. Dropout as a Bayesian approximation: representing model uncertainty in deep learning. ICML 48, 1050\u20131059 (2016).<\/span><\/section>\n
30. Smith, A. M. et al. Standard machine learning approaches outperform deep representation learning on phenotype prediction from transcriptomics data. BMC Bioinformatics 21, 119 (2020).<\/span><\/section>\n
31. Tibshirani, R. Regression shrinkage and selection via the lasso. J. R. Stat. Soc. Ser. B. 58, 267\u2013288<\/span>(1996).<\/span><\/section>\n
32. Zou, H. & Hastie, T. Regularization and variable selection via the elastic net. J. R. Stat. Soc. Ser. B. 67,301\u2013320 (2005).<\/span><\/section>\n
33. Noble, W. S. What is a support vector machine? Nat. Biotechnol. 24, 1565\u20131567 (2006).<\/span><\/section>\n
34. Ben- Hur, A. & Weston, J. A user\u2019s guide to support vector machines. Methods Mol. Biol. 609, 223\u2013239(2010).<\/span><\/section>\n
35. Ben- Hur, A., Ong, C. S., Sonnenburg, S., Sch\u00f6lkopf, B.& R\u00e4tsch, G. Support vector machines and kernels for computational biology. PLoS Comput. Biol. 4,e1000173 (2008). <\/span>This is an introduction to SVMs with a focus on biological data and prediction tasks.<\/span><\/section>\n
36. Kircher, M. et al. A general framework for estimating the relative pathogenicity of human genetic variants.Nat. Genet. 46, 310\u2013315 (2014).<\/span><\/section>\n
37. Driscoll, M. K. et al. Robust and automated detection of subcellular morphological motifs in 3D microscopy images. Nat. Methods 16, 1037\u20131044 (2019).<\/span><\/section>\n
38. Bzdok, D., Krzywinski, M. & Altman, N. Machine learning: supervised methods. Nat. Methods 15, 5\u20136 <\/span>(2018).<\/span><\/section>\n
39. Wang, C. & Zhang, Y. Improving scoring- docking-screening powers of protein- ligand scoring functions using random forest. J. Comput. Chem. 38, 169\u2013177(2017).<\/span><\/section>\n
40. Zeng, W., Wu, M. & Jiang, R. Prediction of enhancer- promoter interactions via natural language processing. BMC Genomics 19, 84 (2018).<\/span><\/section>\n
41. Olson, R. S., Cava, W. L., Mustahsan, Z., Varik, A. & Moore, J. H. Data- driven advice for applying machine learning to bioinformatics problems. Pac. Symp. Biocomput. 23, 192\u2013203 (2018).<\/span><\/section>\n
42. Rappoport, N. & Shamir, R. Multi- omic and multi- view clustering algorithms: review and cancer benchmark.Nucleic Acids Res. 47, 1044 (2019).<\/span><\/section>\n
43. Steinegger, M. & S\u00f6ding, J. MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets. Nat. Biotechnol. 35,1026\u20131028 (2017).<\/span><\/section>\n
44. Jain, A. K. Data clustering: 50 years beyond K- means.Pattern Recognit. Lett. 31, 651\u2013666 (2010).<\/span><\/section>\n
45. Ester M., Kriegel H.-P., Sander J., Xu X. A density- based algorithm for discovering clusters in large spatial databases with noise. KDD\u201896 Proc. Second Int. Conf. Knowl. Discov. Data Mining. 96, 226\u2013231 (1996).<\/span><\/section>\n
46. Nguyen, L. H. & Holmes, S. Ten quick tips for effective dimensionality reduction. PLoS Comput. Biol. 15,e1006907 (2019).<\/span><\/section>\n
47. Moon, K. R. et al. Visualizing structure and transitions in high- dimensional biological data. Nat. Biotechnol.37, 1482\u20131492 (2019).<\/span><\/section>\n
48. van der Maaten, L. & Hinton, G. Visualizing data using t- SNE. J. Mach. Learn. Res. 9, 2579\u20132605 (2008).<\/span><\/section>\n
49. Kobak, D. & Berens, P. The art of using t- SNE for single- cell transcriptomics. Nat. Commun. 10, 5416 <\/span>(2019).<\/span><\/section>\n
This article provides a discussion and tips for using t- SNE as a dimensionality reduction technique on single- cell transcriptomics data.<\/span><\/section>\n
50. Crick, F. The recent excitement about neural networks.Nature 337, 129\u2013132 (1989).<\/span><\/section>\n
51. Geirhos, R. et al. Shortcut learning in deep neural networks. Nat. Mach. Intell. 2, 665\u2013673 (2020). <\/span>This article discusses a common problem in deep learning called \u2018shortcut learning\u2019, where the model uses decision rules that do not transfer to real- world data.<\/span><\/section>\n
52. Qian, N. & Sejnowski, T. J. Predicting the secondary structure of globular proteins using neural network models. J. Mol. Biol. 202, 865\u2013884 (1988).<\/span><\/section>\n
53. deFigueiredo, R. J. et al. Neural- network-based classification of cognitively normal, demented, Alzheimer disease and vascular dementia from single photon emission with computed tomography image data from brain. Proc. Natl Acad. Sci. USA 92, 5530\u20135534 (1995).<\/span><\/section>\n
54. Mayr, A., Klambauer, G., Unterthiner, T. & Hochreiter, S.DeepTox: toxicity prediction using deep learning.Front. Environ. Sci. 3, 80 (2016).<\/span><\/section>\n
55. Yang, J. et al. Improved protein structure prediction using predicted interresidue orientations. Proc. Natl Acad. Sci. USA 117, 1496\u20131503 (2020).<\/span><\/section>\n
56. Xu, J., Mcpartlon, M. & Li, J. Improved protein structure prediction by deep learning irrespective of co-evolution information. Nat. Mach. Intell. 3,601\u2013609 (2021).<\/span><\/section>\n
57. Poplin, R. et al. A universal SNP and small- indel variant caller using deep neural networks.Nat. Biotechnol. 36, 983\u2013987 (2018).<\/span><\/section>\n
58. Fudenberg, G., Kelley, D. R. & Pollard, K. S. Predicting 3D genome folding from DNA sequence with Akita.Nat. Methods 17, 1111\u20131117 (2020).<\/span><\/section>\n
59. Zeng, H., Edwards, M. D., Liu, G. & Gifford, D. K.Convolutional neural network architectures for predicting DNA- protein binding. Bioinformatics 32,i121\u2013i127 (2016).<\/span><\/section>\n
60. Yao, R., Qian, J. & Huang, Q. Deep- learning with synthetic data enables automated picking of cryo- EM particle images of biological macromolecules.Bioinformatics 36, 1252\u20131259 (2020).<\/span><\/section>\n
61. Si, D. et al. Deep learning to predict protein backbone structure from high- resolution cryo- EM density maps.Sci. Rep. 10, 4282 (2020).<\/span><\/section>\n
62. Poplin, R. et al. Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning. Nat. Biomed. Eng. 2, 158\u2013164 (2018).<\/span><\/section>\n
63. AlQuraishi, M. End- to-end differentiable learning of protein structure. Cell Syst. 8, 292\u2013301.e3 (2019).<\/span><\/section>\n
64. Heffernan, R., Yang, Y., Paliwal, K. & Zhou, Y.Capturing non- local interactions by long short- term memory bidirectional recurrent neural networks for improving prediction of protein secondary  structure,backbone angles, contact numbers and solvent accessibility. Bioinformatics 33, 2842\u20132849 (2017).<\/span><\/section>\n
65. M\u00fcller, A. T., Hiss, J. A. & Schneider, G. Recurrent neural network model for constructive peptide design.J. Chem. Inf. Model. 58, 472\u2013479 (2018).<\/span><\/section>\n
66. Choi, E., Bahadori, M. T., Schuetz, A., Stewart, W. F.& Sun, J. Doctor AI: predicting clinical events via recurrent neural networks. JMLR Workshop Conf.Proc. 56, 301\u2013318 (2016).<\/span><\/section>\n
67. Quang, D. & Xie, X. DanQ: a hybrid convolutional and recurrent deep neural network for quantifying the function of DNA sequences. Nucleic Acids Res. 44,e107 (2016).<\/span><\/section>\n
68. Alley, E. C., Khimulya, G., Biswas, S., AlQuraishi, M.& Church, G. M. Unified rational protein engineering with sequence- based deep representation learning.Nat. Methods 16, 1315\u20131322 (2019).<\/span><\/section>\n
69. Vaswani, A. et al. Attention is all you need.arXiv <\/span>https:\/\/arxiv.org\/abs\/1706.03762 (2017).<\/span><\/span><\/section>\n
70. Elnaggar, A. et al. ProtTrans: towards cracking the language of life\u2019s code through self-supervised deep learning and high performance computing. arXiv <\/span>https:\/\/arxiv.org\/abs\/2007.06225 (2020).<\/span><\/span><\/section>\n
71. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583\u2013589 (2021).<\/span><\/section>\n
72. Battaglia, P. W. et al. Relational inductive biases, deep learning, and graph networks. arXiv <\/span>https:\/\/arxiv.org\/ abs\/1806.01261<\/span><\/span> (2018).<\/span><\/section>\n
73. Stokes, J. M. et al. A deep learning approach to antibiotic discovery. Cell 181, 475\u2013483 (2020). <\/span>In this work, a deep learning model predicts antibiotic activity, with one candidate showing broad- spectrum antibiotic activities in mice.<\/span><\/section>\n
74. Gainza, P. et al. Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning. Nat. Methods 17, 184\u2013192 (2020).<\/span><\/section>\n
75. Strokach, A., Becerra, D., Corbi- Verge, C., Perez- Riba, A.& Kim, P. M. Fast and flexible protein design using deep graph neural networks. Cell Syst. 11, 402\u2013411.e4(2020).<\/span><\/section>\n
76. Gligorijevic, V. et al. Structure-based function prediction using graph convolutional networks. Nat.   Commun. 12, 3168 (2021).<\/span><\/section>\n
77. Zitnik, M., Agrawal, M. & Leskovec, J. Modeling polypharmacy side effects with graph convolutional networks. Bioinformatics 34, i457\u2013i466 (2018).<\/span><\/section>\n
78. Veselkov, K. et al. HyperFoods: machine intelligent mapping of cancer- beating molecules in foods.Sci. Rep. 9, 9237 (2019).<\/span><\/section>\n
79. Fey, M. & Lenssen, J. E. Fast graph representation learning with PyTorch geometric. arXiv <\/span>https:\/\/arxiv.org\/abs\/1903.02428 (2019).<\/span><\/span><\/section>\n
80. Zhavoronkov, A. et al. Deep learning enables rapid identification of potent DDR1 kinase inhibitors. Nat. Biotechnol. 37, 1038\u20131040 (2019).<\/span><\/section>\n
81. Wang, Y. et al. Predicting DNA methylation state of CpG dinucleotide using genome topological features and deep networks. Sci. Rep. 6, 19598(2016).<\/span><\/section>\n
82. Linder, J., Bogard, N., Rosenberg, A. B. & Seelig, G.A generative neural network for maximizing fitness and diversity of synthetic DNA and protein sequences.Cell Syst. 11, 49\u201362.e16 (2020).<\/span><\/section>\n
83. Greener, J. G., Moffat, L. & Jones, D. T. Design of metalloproteins and novel protein folds using variational autoencoders. Sci. Rep. 8, 16189(2018).<\/span><\/section>\n
84. Wang, J. et al. scGNN is a novel graph neural network framework for single- cell RNA- Seq analyses. Nat. Commun. 12, 1882 (2021).<\/span><\/section>\n
85. Paszke, A. et al. PyTorch: an imperative style, high- performance deep learning library. Adv. Neural Inf. Process. Syst. 32, 8024\u20138035 (2019).<\/span><\/section>\n
86. Abadi M. et al. Tensorflow: a system for large- scale machine learning. 12th USENIX Symposium on Operating Systems Design and Implementation.265\u2013283 (USENIX, 2016).<\/span><\/section>\n
87. Wei, Q. & Dunbrack, R. L. Jr The role of balanced training and testing data sets for binary classifiers in bioinformatics. PLoS ONE 8, e67863 (2013).<\/span><\/section>\n
88. Walsh, I., Pollastri, G. & Tosatto, S. C. E. Correct machine learning on protein sequences: a peer-reviewing perspective. Brief. Bioinform 17, 831\u2013840(2016).<\/span><\/section>\n
This article discusses how peer reviewers can assess machine learning methods in biology, and by extension how scientists can design and conduct such studies properly.<\/span><\/section>\n
89. Schreiber, J., Singh, R., Bilmes, J. & Noble, W. S.A pitfall for machine learning methods aiming to predict across cell types. Genome Biol. 21, 282(2020).<\/span><\/section>\n
90. Chothia, C. & Lesk, A. M. The relation between the divergence of sequence and structure in proteins. EMBO J. 5, 823\u2013826 (1986).<\/span><\/section>\n
91. S\u00f6ding, J. & Remmert, M. Protein sequence comparison and fold recognition: progress and good- practice benchmarking. Curr. Opin. Struct. Biol.21, 404\u2013411 (2011).<\/span><\/section>\n
92. Steinegger, M. et al. HH- suite3 for fast remote homology detection and deep protein annotation. BMC Bioinformatics 20, 473 (2019).<\/span><\/section>\n
93. Sillitoe, I. et al. CATH: expanding the horizons of structure- based functional annotations for genome sequences. Nucleic Acids Res. 47, D280\u2013D284(2019).<\/span><\/section>\n
94. Cheng, H. et al. ECOD: an evolutionary classification of protein domains. PLoS Comput. Biol. 10, e1003926(2014).<\/span><\/section>\n
95. Li, Y. & Yang, J. Structural and sequence similarity makes a significant impact on machine- learning-based scoring functions for protein- ligand interactions.J. Chem. Inf. Model. 57, 1007\u20131012 (2017).<\/span><\/section>\n
96. Zech, J. R. et al. Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: a cross- sectional study. PLoS Med. 15,e1002683 (2018).<\/span><\/section>\n
97. Szegedy, C. et al. Intriguing properties of neural networks. arXiv <\/span>https:\/\/arxiv.org\/abs\/1312.6199 (2014).<\/span><\/span><\/section>\n
98. Hie, B., Cho, H. & Berger, B. Realizing private and practical pharmacological collaboration. Science 362,347\u2013350 (2018).<\/span><\/section>\n
99. Beaulieu- Jones, B. K. et al. Privacy- preserving generative deep neural networks support clinical data sharing. Circ. Cardiovasc. Qual. Outcomes 12,e005122 (2019).<\/span><\/section>\n
100. Kone\u010dn\u00fd, J., Brendan McMahan, H., Ramage, D.& Richt\u00e1rik, P. Federated optimization: distributed machine learning for on-device intelligence. arXiv <\/span>https:\/\/arxiv.org\/abs\/1610.02527 (2016).<\/span><\/span><\/section>\n
101. P\u00e9rez, A., Mart\u00ednez- Rosell, G. & De Fabritiis, G.Simulations meet machine learning in structural biology. Curr. Opin. Struct. Biol. 49, 139\u2013144 (2018).<\/span><\/section>\n
102. No\u00e9, F., Olsson, S., K\u00f6hler, J. & Wu, H. Boltzmann generators: sampling equilibrium states of many-body systems with deep learning. Science 365, 6457(2019).<\/span><\/section>\n
103. Shrikumar, A., Greenside, P. & Kundaje, A. Reverse-complement parameter sharing improves deep learning models for genomics. bioRxiv <\/span>https:\/\/www.biorxiv.org\/content\/10.1101\/103663v1 (2017).<\/span><\/span><\/section>\n
104. Lopez, R., Gayoso, A. & Yosef, N. Enhancing scientific discoveries in molecular biology with deep generative models. Mol. Syst. Biol. 16, e9198 (2020).<\/span><\/section>\n
105. Anishchenko, I., Chidyausiku, T. M., Ovchinnikov, S.,Pellock, S. J. & Baker, D. De novo protein design by deep network hallucination. bioRxiv <\/span>https:\/\/doi.org\/10.1101\/2020.07.22.211482 (2020).<\/span><\/span><\/section>\n
106. Innes, M. et al. A differentiable programming system to bridge machine learning and scientific computing. arXiv   <\/span>https:\/\/arxiv.org\/abs\/1907.07587 (2019).<\/span><\/span><\/section>\n
107. Ingraham J., Riesselman A. J., Sander C., Marks D. S.Learning protein structure with a differentiable simulator.ICLR <\/span>https:\/\/openreview.net\/forum?id=Byg3y3C9Km(2019).<\/span><\/section>\n
108. Jumper, J. M., Faruk, N. F., Freed, K. F. & Sosnick, T. R.Trajectory- based training enables protein simulations with accurate folding and Boltzmann ensembles in cpu- hours. PLoS Comput. Biol. 14, e1006578 (2018).<\/span><\/section>\n
109. Wang, Y., Fass, J. &<\/span> Chodera, J. D. End-to-end differentiable molecular mechanics force field construction. arXiv <\/span>http:\/\/arxiv.org\/abs\/2010.01196(2020).<\/span><\/section>\n
110. Bradbury, J. et al. JAX: composable transformations of Python+NumPy programs. GitHub <\/span>http:\/\/github.com\/google\/jax (2018).<\/span><\/span><\/section>\n
111. Chen, K. M., Cofer, E. M., Zhou, J. & Troyanskaya, O. G.Selene: a PyTorch- based deep learning library for sequence data. Nat. Methods 16, 315\u2013318 (2019).<\/span><\/section>\n
This work provides a software library based on PyTorch providing functionality for biological sequences.<\/span><\/section>\n
112. Kopp, W., Monti, R., Tamburrini, A., Ohler, U.& Akalin, A. Deep learning for genomics using Janggu. Nat. Commun. 11, 3488 (2020).<\/span><\/section>\n
113. Schoenholz, S. S. & Cub<\/span>uk, E. D. JAX, M.D.: end-to-end differentiable, hardware accelerated, molecular dynamics in pure Python. arXiv <\/span>https:\/\/arxiv.org\/abs\/1912.04232 (2019).<\/span><\/span><\/section>\n
114. Avsec, \u017d. et al. The Kipoi repository accelerates community exchange and reuse of predictive models for genomics. Nat. Biotechnol. 37, 592\u2013600 (2019).<\/span><\/section>\n
115. Isensee, F., Jaeger, P. F., Kohl, S. A. A., Petersen, J.& Maier- Hein, K. H. nnU- Net: a self- configuring method for deep learning- based biomedical image segmentation. Nat Methods 18, 203\u2013211 (2020).<\/span><\/section>\n
116. Livesey, B. J. & Marsh, J. A. Using deep mutational scanning to benchmark variant effect predictors and identify disease mutations. Mol. Syst. Biol. 16, e9380 (2020).<\/span><\/section>\n
117. AlQuraishi, M. ProteinNet: a standardized data set for machine learning of protein structure.BMC Bioinformatics 20, 311 (2019).<\/span><\/section>\n
118. Townshend, R. J. L. et al. ATOM3D: tasks on molecules in three dimensions. arXiv  <\/span>https:\/\/arxiv.org\/abs\/2012.04035 (2020).<\/span><\/span><\/section>\n
119. Rao, R. et al. Evaluating protein transfer learning with TAPE. Adv. Neural. Inf. Process. Syst. 32, 9689\u20139701(2019).<\/span><\/section>\n
120. Kryshtafovych, A., Schwede, T., Topf, M., Fidelis, K.& Moult, J. Critical assessment of methods of protein structure prediction (CASP) \u2014 round XIII. Proteins 87,1011\u20131020 (2019).<\/span><\/section>\n
121. Zhou, N. et al. The CAFA challenge reports improved protein function prediction and new functional annotations for hundreds of genes through experimental screens. Genome Biol. 20, 244 (2019).<\/span><\/section>\n
122. Munro, D. & Singh, M. DeMaSk: a deep mutational scanning substitution matrix and its use for variant impact prediction. Bioinformatics 36, 5322\u20135329(2020).<\/span><\/section>\n
123. Haario, H. & Taavitsainen, V.-M. Combining soft and hard modelling in chemical kinetic models. Chemom.Intell. Lab. Syst. 44, 77\u201398 (1998).<\/span><\/section>\n
124. Cozzetto, D., Minneci, F., Currant, H. & Jones, D. T.FFPred 3: feature- based function prediction for all gene ontology domains. Sci. Rep. 6, 31865 (2016).<\/span><\/section>\n
125. Nugent, T. & Jones, D. T. Transmembrane protein topology prediction using support vector machines.BMC Bioinformatics 10, 159 (2009).<\/span><\/section>\n
126. Bao, L., Zhou, M. & Cui, Y. nsSNPAnalyzer: identifying disease- associated nonsynonymous single nucleotide polymorphisms. Nucleic Acids Res. 33, W480\u2013W482(2005).<\/span><\/section>\n
127. Li, W., Yin, Y., Quan, X. & Zhang, H. Gene expression value prediction based on XGBoost algorithm. Front.Genet. 10, 1077 (2019).<\/span><\/section>\n
128. Zhang, Y. & Skolnick, J. SPICKER: a clustering approach to identify near- native protein folds. J. Comput. Chem.30, 865\u2013871 (2004).<\/span><\/section>\n
129. Teodoro, M. L., Phillips, G. N. Jr & Kavraki, L. E.Understanding protein flexibility through dimensionality reduction. J. Comput. Biol. 10,617\u2013634 (2003).<\/span><\/section>\n
130. Schlichtkrull, M. et al. Modeling relational data with graph convolutional networks. arXiv <\/span>https:\/\/arxiv.org\/abs\/1703.06103 (2019).<\/span><\/span><\/section>\n
131. Pandarinath, C. et al. Inferring single- trial neural population dynamics using sequential auto- encoders.Nat. Methods 15, 805\u2013815 (2018).<\/span><\/section>\n
132. Antczak, M., Michaelis, M. & Wass, M. N.Environmental conditions shape the nature of a minimal bacterial genome. Nat. Commun. 10, 3100 (2019).<\/span><\/section>\n
133. Sun, T., Zhou, B., Lai, L. & Pei, J. Sequence- based prediction of protein protein interaction using a deep- learning algorithm. BMC Bioinformatics 18,277 (2017).<\/span><\/section>\n
134. Hiranuma, N. et al. Improved protein structure refinement guided by deep learning based accuracy estimation. Nat. Commun. 12, 1340 (2021).<\/span><\/section>\n
135. Pag\u00e8s, G., Charmettant, B. & Grudinin, S. Protein model quality assessment using 3D oriented convolutional neural networks. Bioinformatics 35,3313\u20133319 (2019).<\/span><\/section>\n
136. Pires, D. E. V., Ascher, D. B. & Blundell, T. L. DUET: a server for predicting effects of mutations on protein stability using an integrated computational approach.Nucleic Acids Res. 42, W314\u2013W319 (2014).<\/span><\/section>\n
137. Yuan, Y. & Bar- Joseph, Z. Deep learning for inferring gene relationships from single- cell expression data.Proc. Natl Acad. Sci. USA 116, 27151\u201327158 (2019).<\/span><\/section>\n
138. Chen, L., Cai, C., Chen, V. & Lu, X. Learning a hierarchical representation of the yeast transcriptomic machinery using an autoencoder model. BMC Bioinformatics 17, S9 (2016).<\/span><\/section>\n
139. Kantz, E. D., Tiwari, S., Watrous, J. D., Cheng, S.& Jain, M. Deep neural networks for classification of LC- MS spectral peaks. Anal. Chem. 91, 12407\u201312413(2019).<\/span><\/section>\n
140. D\u00fchrkop, K. et al. SIRIUS 4: a rapid tool for turning tandem mass spectra into metabolite structure information. Nat. Methods 16, 299\u2013302 (2019).<\/span><\/section>\n
141. Liebal, U. W., Phan, A. N. T., Sudhakar, M., Raman, K. & Blank, L. M. Machine learning applications for mass spectrometry- based metabolomics. Metabolites 10,243 (2020).<\/span><\/section>\n
142. Zhong, E. D., Bepler, T., Berger, B. & Davis, J. H. CryoDRGN: reconstruction of heterogeneous cryo- EM structures using neural networks. Nat. Methods 18,176\u2013185 (2021).<\/span><\/section>\n
143. Schmauch, B. et al. A deep learning model to predict RNA- Seq expression of tumours from whole slide images. Nat. Commun. 11, 3877 (2020).<\/span><\/section>\n
144. Das, P. et al. Accelerated antimicrobial discovery via deep generative models and molecular dynamics simulations. Nat. Biomed. Eng. 5, 613\u2013623 (2021).<\/span><\/section>\n
145. Gligorijevic, <\/span>V., Barot, M. & Bonneau, R. deepNF:deep network fusion for protein function prediction.Bioinformatics 34, 3873\u20133881 (2018).<\/span><\/section>\n
146. Karpathy A. A recipe for training neural networks. <\/span>https:\/\/karpathy.github.io\/2019\/04\/25\/recipe (2019).<\/span><\/span><\/section>\n
147. Bengio, Y. Practical recommendations for gradient-based training of deep architectures. Lecture Notes Comput. Sci. 7700, 437\u2013478 (2012).<\/span><\/section>\n
148. Roberts, M. et al. Common pitfalls and recommendations for using machine learning to detect and prognosticate for COVID-19 using chest radiographs and CT scans. Nat. Mach. Intell. 3,199\u201321<\/span>7 (2021).<\/span><\/section>\n
This study assesses 62 machine learning studies that analyse medical images for COVID-19 and none is found to be of clinical use, indicating the difficulties of training a useful model.<\/span><\/section>\n
149. List, M., Ebert, P. & Albrecht, F. Ten simple rules for developing usable software in computational biology.PLoS Comput. Biol. 13, e1005265 (2017).<\/span><\/section>\n
150. Sonnenburg, S. \u00c3., Braun, M. L., Ong, C. S. & Bengio, S.The need for open source software in machine learning.J. Mach. Learn. Res. 8, 2443\u20132466 (2007).<\/span><\/section>\n


<\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\n
\n

\uff08\u53c2\u8003\u6587\u732e\u53ef\u4e0a\u4e0b\u6ed1\u52a8<\/strong>\u67e5\u770b\uff09<\/p>\n<\/section>\n<\/section>\n


<\/section>\n

<\/section>\n
\n
\n
\n
\n

\u590d\u6742\u79d1\u5b66\u6700\u65b0\u8bba\u6587<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\n


<\/p>\n<\/section>\n

\u96c6\u667a\u6591\u56fe\u9876\u520a\u8bba\u6587\u901f\u9012\u680f\u76ee\u4e0a\u7ebf\u4ee5\u6765\uff0c\u6301\u7eed\u6536\u5f55\u6765\u81eaNature\u3001Science<\/span>\u7b49\u9876\u520a\u7684\u6700\u65b0\u8bba\u6587\uff0c\u8ffd\u8e2a\u590d\u6742\u7cfb\u7edf\u3001\u7f51\u7edc\u79d1\u5b66\u3001\u8ba1\u7b97\u793e\u4f1a\u79d1\u5b66\u7b49\u9886\u57df\u7684\u524d\u6cbf\u8fdb\u5c55\u3002\u73b0\u5728\u6b63\u5f0f\u63a8\u51fa\u8ba2\u9605\u529f\u80fd\uff0c\u6bcf\u5468\u901a\u8fc7\u5fae\u4fe1\u670d\u52a1\u53f7\u300c\u96c6\u667a\u6591\u56fe\u300d\u63a8\u9001\u8bba\u6587\u4fe1\u606f\u3002\u626b\u63cf\u4e0b\u65b9\u4e8c\u7ef4\u7801\u5373\u53ef\u4e00\u952e\u8ba2\u9605\uff1a<\/span>
<\/p>\n

<\/p>\n

\n


<\/p>\n<\/section>\n


<\/p>\n

\n

\u63a8\u8350\u9605\u8bfb<\/span><\/strong><\/strong><\/p>\n