{"id":54091,"date":"2024-12-07T19:15:28","date_gmt":"2024-12-07T11:15:28","guid":{"rendered":"https:\/\/swarma.org\/?p=54091"},"modified":"2024-12-07T19:15:28","modified_gmt":"2024-12-07T11:15:28","slug":"%e6%8e%a7%e5%88%b6%e7%a7%91%e5%ad%a6-x-%e6%9c%ba%e5%99%a8%e5%ad%a6%e4%b9%a0%e5%89%8d%e6%b2%bf%ef%bc%9apid%e6%8e%a7%e5%88%b6%e5%99%a8%e5%8a%a0%e9%80%9f%e4%ba%ba%e5%b7%a5%e6%99%ba%e8%83%bd%e4%bc%98","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=54091","title":{"rendered":"\u63a7\u5236\u79d1\u5b66 x \u673a\u5668\u5b66\u4e60\u524d\u6cbf\uff1aPID\u63a7\u5236\u5668\u52a0\u901f\u4eba\u5de5\u667a\u80fd\u4f18\u5316\u8bad\u7ec3\uff5cNat. Commun.\u901f\u9012"},"content":{"rendered":"
\n
\n
\n
\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

\u8fd1\u65e5\uff0cNature Communications<\/em> \u53d1\u8868\u9898\u4e3a\u201cAccelerated optimization in deep learning with a proportional-integral derivative controller\u201d\u7684\u6700\u65b0\u8bba\u6587\uff0c\u8be5\u7814\u7a76\u5c06\u95ed\u73af\u53cd\u9988\u63a7\u5236\u4e0e\u4f18\u5316\u7b97\u6cd5\u5efa\u7acb\u8054\u7cfb\uff0c\u521b\u65b0\u6027\u5730\u63d0\u51fa\u9762\u5411\u6df1\u5ea6\u5b66\u4e60\u7684\u53d7\u63a7\u52a0\u901f\u4f18\u5316\u7b97\u6cd5 PIDAO\uff08\u6ce8\uff1aPIDAO \u8bfb\u4f5c\u201c\u8f9f\u9053\u201d\uff09\u3002\u672c\u7814\u7a76\u4e0d\u4ec5\u6709\u52a9\u4e8e\u52a0\u901f\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u7684\u8bad\u7ec3\uff0c\u63d0\u51fa\u7684\u865a\u62df\u7269\u7406\u5b66\u4e0e\u63a7\u5236\u7406\u8bba\u7b49\u5b66\u79d1\u4ea4\u53c9\u65b9\u6cd5\u4e5f\u4e3a\u6df1\u5ea6\u5b66\u4e60\u7814\u7a76\u53e6\u8f9f\u8e4a\u5f84\u3002<\/span><\/strong><\/span><\/span><\/span><\/strong><\/p>\n<\/section>\n

\n


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

\u5173\u952e\u8bcd\uff1a\u6df1\u5ea6\u5b66\u4e60\uff0c\u63a7\u5236\u7406\u8bba\uff0c\u52a8\u529b\u7cfb\u7edf\uff0c\u865a\u62df\u7269\u7406\u5b66\uff0cControl for Learning<\/span><\/strong><\/span><\/strong><\/strong><\/span><\/strong><\/strong><\/strong><\/section>\n
<\/strong><\/strong><\/section>\n
<\/strong><\/strong><\/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
<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

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

<\/p>\n

\n

\u8bba\u6587\u9898\u76ee\uff1aAccelerated optimization in deep learning  with a proportional-integral derivative controlle<\/span><\/p>\n

\u8bba\u6587\u5730\u5740\uff1ahttps:\/\/www.nature.com\/articles\/s41467-024-54451-3<\/span><\/p>\n

\u8bba\u6587\u4f5c\u8005\uff1a\u9648\u5d69\u3001\u5218\u4f73\u65ed\u3001\u738b\u9e4f\u51ef\u3001\u8bb8\u8d85\u3001\u8521\u58f0\u6cfd\u3001\u891a\u5065<\/span><\/span><\/p>\n<\/blockquote>\n

\n


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

\u4eba\u5de5\u667a\u80fd\u5df2\u5728\u4f17\u591a\u9886\u57df\u5c55\u73b0\u4e86\u5353\u8d8a\u7684\u5e94\u7528\u6f5c\u529b\uff0c\u4f46AI\u6a21\u578b\u7684\u8bad\u7ec3\u5f80\u5f80\u9700\u8981\u8017\u8d39\u5927\u91cf\u65f6\u95f4\u548c\u8ba1\u7b97\u8d44\u6e90\uff0c\u80cc\u540e\u4e5f\u8574\u542b\u7740\u590d\u6742\u7684\u4f18\u5316\u7406\u8bba\u548c\u52a8\u529b\u5b66\u95ee\u9898\u3002\u6a21\u578b\u7684\u8bad\u7ec3<\/span>\uff08\u5373\u76ee\u6807\u51fd\u6570\u7684\u4f18\u5316\uff09<\/span>\u95ee\u9898\u901a\u5e38\u9762\u4e34\u4e24\u4e2a\u4e3b\u8981\u6311\u6218\uff1a\u4e00\u662f\u9ad8\u7ef4\u975e\u51f8\u4f18\u5316\u95ee\u9898\u5e26\u6765\u7684\u590d\u6742\u635f\u5931\u666f\u89c2\uff1b\u4e8c\u662f\u5927\u89c4\u6a21\u6570\u636e\u548c\u6df1\u5c42\u7f51\u7edc\u7684\u8ba1\u7b97\u4ee3\u4ef7\u3002\u5728\u672c\u9879\u5de5\u4f5c\u4e2d\uff0c\u7814\u7a76\u56e2\u961f\u878d\u5408\u6570\u503c\u4f18\u5316\u3001\u52a8\u529b\u7cfb\u7edf\u3001\u63a7\u5236\u7406\u8bba\u3001\u6570\u503c\u8ba1\u7b97\u7b49\u591a\u4e2a\u65b9\u5411\uff0c\u4e3a\u7406\u89e3\u6df1\u5ea6\u5b66\u4e60\u7684\u8bad\u7ec3\u673a\u5236\u5e26\u6765\u65b0\u7684\u7814\u7a76\u89c6\u89d2<\/strong>\u3002
<\/span><\/p>\n


<\/p>\n

\u5f53\u524d\uff0c\u591a\u6570\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u7684\u8bad\u7ec3\u6700\u7ec8\u53ef\u5f52\u7ed3\u4e3a\u9ad8\u7ef4\u4f18\u5316\u95ee\u9898\u6c42\u89e3\uff0c\u800c\u57fa\u4e8e\u68af\u5ea6\u7684\u65b9\u6cd5\u662f\u6c42\u89e3\u4f18\u5316\u95ee\u9898\u7684\u5173\u952e\uff0c\u5982SGD\u3001Adam\u3001\u91cd\u7403\u6cd5\u7b49\u5e38\u7528\u65b9\u6cd5\u3002\u5982\u4f55\u7406\u89e3\u8fd9\u4e9b\u4f18\u5316\u7b97\u6cd5\u7684\u672c\u8d28\u53ca\u5176\u52a8\u6001\u8fc7\u7a0b\uff0c\u662f\u7a81\u7834\u6df1\u5ea6\u5b66\u4e60\u53ef\u89e3\u91ca\u6027\u7684\u4e00\u5927\u6311\u6218\u3002\u56e2\u961f\u901a\u8fc7\u5f15\u5165\u865a\u62df\u7269\u7406\u5b66\u6765\u5efa\u6a21\u4f18\u5316\u8fc7\u7a0b<\/span>\uff08\u5982\u5f53\u524d\u5de5\u4f5c\u8fd0\u7528\u4e86\u7ecf\u5178\u529b\u5b66\u4e8c\u9636ODE\u5efa\u6a21\uff09<\/span>\uff0c\u968f\u540e\u628a\u4f18\u5316\u95ee\u9898\u8f6c\u5316\u6210\u9ad8\u7ef4\u53c2\u6570\u7a7a\u95f4\u4e2d\u7684\u52a8\u529b\u5b66\u63a7\u5236\u95ee\u9898\uff0c\u5982\u6b64\u4e00\u6765\uff0c\u4e30\u5bcc\u7684\u53cd\u9988\u63a7\u5236\u7406\u8bba\u4f53\u7cfb\u53ef\u4ee5\u4e3a\u4f18\u5316\u7b97\u6cd5\u8bbe\u8ba1\u4e0e\u6536\u655b\u6027\u5206\u6790\u63d0\u4f9b\u5168\u65b0\u7684\u601d\u8def<\/strong>\u3002 
<\/span><\/p>\n


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

\n

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

\n
\u91cd\u7403\u6cd5\uff08Heavy-ball optimizer\uff09\u7684\u52a8\u529b\u7cfb\u7edf\u89c2\u70b9\uff0c\u5229\u7528Lyapunov\u7a33\u5b9a\u6027\u5206\u6790\u7b97\u6cd5\u6536\u655b\u6027<\/span><\/span><\/section>\n<\/section>\n
\n


<\/p>\n

\u9488\u5bf9\u5982\u4f55\u52a0\u901f\u8bad\u7ec3\u795e\u7ecf\u7f51\u7edc\u8fd9\u4e00\u95ee\u9898\uff0c\u8be5\u5de5\u4f5c\u5c06\u63a7\u5236\u79d1\u5b66\u4e2d\u4f7f\u7528\u6700\u4e3a\u5e7f\u6cdb\u7684\u6bd4\u4f8b-\u79ef\u5206-\u5fae\u5206<\/span>\uff08PID\uff09<\/span>\u53cd\u9988\u63a7\u5236\u5f8b\u5e94\u7528\u4e8e\u523b\u753b\u4f18\u5316\u8fc7\u7a0b\u7684\u865a\u62df\u7269\u7406\u52a8\u529b\u5b66\uff0c\u521b\u65b0\u5730\u63d0\u51fa\u7528\u4e8e\u6df1\u5ea6\u5b66\u4e60\u7684\u8fde\u7eed\u65f6\u95f4PID\u52a0\u901f\u4f18\u5316\u5668<\/span>\uff08Proportional-Integral-Derivative Accelerated Optimization, PIDAO\uff09<\/span>\u3002\u901a\u8fc7\u6784\u5efa\u5408\u9002\u7684Lyapunov\u51fd\u6570\u6765\u5efa\u7acbPIDAO\u4e2d\u8d85\u53c2\u6570\u4e0e\u7a33\u5b9a\u6027\u4e4b\u95f4\u7684\u5173\u7cfb\uff0c\u4ee5\u53ca\u52a8\u529b\u7cfb\u7edf\u5e73\u8861\u70b9\u4e0e\u9ad8\u7ef4\u53c2\u6570\u7a7a\u95f4\u8bad\u7ec3\u95ee\u9898\u6700\u4f18\u6027\u7684\u5173\u8054\u3002 <\/span><\/p>\n


<\/p>\n

\u76f8\u6bd4\u4e8ePolyak\u63d0\u51fa\u7684\u91cd\u7403\u6cd5\uff0cPIDAO\u4e0d\u518d\u5355\u4e00\u53d7\u68af\u5ea6\u8c03\u63a7\u5176\u4f18\u5316\u65b9\u5411\uff0c\u800c\u662f\u901a\u8fc7PID\u53cd\u9988\u7ed9\u5bfb\u4f18\u673a\u5668\u4eba\u4ee5\u51fd\u6570\u68af\u5ea6<\/span>\uff08\u5f53\u524d\u201c\u5730\u52bf\u201d\uff09<\/span>\u3001\u4e00\u6bb5\u5386\u53f2\u68af\u5ea6\u79ef\u5206<\/span>\uff08\u8fc7\u5f80\u201c\u5730\u52bf\u201d\uff09<\/span>\u4ee5\u53ca\u524d\u65b9\u68af\u5ea6\u8d8b\u52bf<\/span>\uff08\u672a\u6765\u201c\u5730\u52bf\u201d\uff09<\/span>\u4e09\u7c7b\u4fe1\u606f\uff0c\u6765\u63a7\u5236\u5bfc\u822a\u5bfb\u4f18\u8fc7\u7a0b\u3002\u7b80\u5355\u6765\u8bf4\uff0cPolyak\u65b9\u6cd5\u7528\u5230\u4e86\u201c\u73b0\u5728\u201d\uff0cNesterov\u65b9\u6cd5\u7528\u5230\u4e86\u201c\u73b0\u5728\u201d\u548c\u201c\u672a\u6765\u201d\uff0c\u800cPIDAO\u878d\u5408\u4e09\u7c7b\u4fe1\u606f\u5171\u540c\u51b3\u7b56\u5bfb\u4f18\u3002 <\/span><\/p>\n


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

\n

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

\n

\u7528\u4e8e\u6df1\u5ea6\u5b66\u4e60\u7684PID\u52a0\u901f\u4f18\u5316\u7b97\u6cd5\u7684\u57fa\u672c\u601d\u60f3<\/span><\/p>\n<\/section>\n


<\/p>\n

\n

\u6b64\u5916\uff0c\u57fa\u4e8eLyapunov\u7a33\u5b9a\u6027\u7406\u8bba\uff0c\u7814\u7a76\u56e2\u961f\u63d0\u4f9b\u4e86PIDAO\u5728\u51f8\u4f18\u5316\u95ee\u9898\u548c\u975e\u51f8\u4f18\u5316\u95ee\u9898\u7684\u8d85\u53c2\u6570\u9009\u53d6\u8303\u56f4\uff0c\u540c\u65f6\u53d1\u73b0PIDAO\u4f18\u5316\u5668\u7684\u62b5\u6d88\u6548\u5e94\uff0c\u8be5\u7279\u6027\u4f7f\u5f97PIDAO\u5177\u6709\u66f4\u5f3a\u7684\u63a2\u7d22\u635f\u5931\u666f\u89c2\u7684\u80fd\u529b\u3002\u6b64\u5916\uff0c\u501f\u52a9 PID \u63a7\u5236\u5668\u548c PIDAO \u4e4b\u95f4\u7684\u8054\u7cfb\uff0c\u7814\u7a76\u56e2\u961f\u63d0\u4f9b\u4e86PIDAO\u4f18\u5316\u5668\u7684\u8d85\u53c2\u6570\u53ef\u89e3\u91ca\u6027\u3002\u5229\u7528\u4e0d\u540c\u79bb\u6563\u5316\u65b9\u6cd5\u5bf9PIDAO\u79bb\u6563\u5316\uff0c\u8fdb\u800c\u53d1\u5c55\u51fa\u7528\u4e8e\u795e\u7ecf\u7f51\u7edc\u8bad\u7ec3\u7684PIDAO\u4f18\u5316\u5668\u5e76\u96c6\u6210\u4e8e\u6df1\u5ea6\u5b66\u4e60\u6846\u67b6<\/span>\uff08\u5982Pytorch\uff09<\/span>\u3002\u4e0e\u57fa\u51c6\u4f18\u5316\u7b97\u6cd5\u65b9\u6cd5<\/span>\uff08\u5982Adam\uff09<\/span>\u76f8\u6bd4\uff0cPIDAO\u7cfb\u5217\u7b97\u6cd5\u5728\u4e0d\u540c\u7684\u6d4b\u8bd5\u573a\u666f\u4e2d\u5747\u83b7\u5f97\u66f4\u9ad8\u7684\u8bad\u7ec3\u7cbe\u5ea6\u548c\u66f4\u5feb\u7684\u6536\u655b\u901f\u5ea6\u3002 <\/span><\/p>\n


<\/p>\n

\u8be5\u9879\u7814\u7a76\u6240\u63d0\u51fa\u7684\u53d7\u63a7\u91cd\u7403\u4f18\u5316\u5668\u6846\u67b6\u548c\u52a0\u901f\u4f18\u5316\u7b97\u6cd5PIDAO\uff0c\u6709\u52a9\u4e8e\u52a0\u901f\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u8bad\u7ec3\uff0c\u540c\u65f6\u4e5f\u4e3a\u57fa\u4e8e\u63a7\u5236\u7406\u8bba\u65b9\u6cd5\u7814\u7a76\u6df1\u5ea6\u5b66\u4e60\u95ee\u9898\u63d0\u4f9b\u4e86\u65b0\u601d\u8def\u3002<\/span><\/p>\n


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

\u8fd1\u5e74\u6765\uff0c\u6d59\u6c5f\u5927\u5b66\u63a7\u5236\u79d1\u5b66\u4e0e\u5de5\u7a0b\u5b66\u9662\u7814\u7a76\u56e2\u961f\u91cd\u70b9\u5173\u6ce8\u63a7\u5236\u79d1\u5b66\u4e0e\u673a\u5668\u5b66\u4e60\u7684\u4ea4\u53c9\u7814\u7a76\uff0c\u7279\u522b\u662f Control for Learning<\/strong><\/span>\uff08\u63a7\u5236\u8d4b\u80fd\u5b66\u4e60\uff09<\/span>\uff0c\u8fd9\u5c5e\u4e8e Science for AI \u8303\u7574\uff0c\u8fd1\u671f\u76f8\u5173\u7684\u5de5\u4f5c\u5305\u62ec\uff1a\u5206\u5e03\u5f0f\u4f18\u5316\u6c42\u89e3<\/span>\uff08Liu et al., IET CSR, 2023\uff09<\/span>\u3001\u57fa\u4e8e\u9c81\u68d2\u63a7\u5236\u7684\u795e\u7ecf\u7f51\u7edc\u8bbe\u8ba1<\/span>\uff08Chen et al., IEEE TNNLS\uff09<\/span>\u3001\u57fa\u4e8e\u53cd\u9988\u7684\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u8bbe\u8ba1<\/span>\uff08Wang et al., Neurocomputing, 2024\uff09<\/span>\u7b49\u3002\u6b64\u5916\uff0c\u56e2\u961f\u5728 AI for Science \u65b9\u9762\u53d1\u8868\u4e86\u5927\u89c4\u6a21\u7c92\u5b50\u667a\u80fd\u53ef\u89c6\u8ba1\u7b97<\/span>\uff08Liang et al., Nature Machine Intelligence\uff09<\/span>\u7b49\u5de5\u4f5c\u3002<\/span><\/p>\n

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

\n

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

\n

\u7ecf\u5178\u7b97\u6cd5\u4e0ePIDAO\u7b97\u6cd5\uff0c\u5728\u51f8\u4f18\u5316\u4e0e\u975e\u51f8\u4f18\u5316\u95ee\u9898\u7684\u5bf9\u6bd4<\/span><\/span><\/p>\n<\/section>\n


<\/p>\n

\n

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

\n

 <\/span>PIDAO\u4e0eAdam\u7b49\u7ecf\u5178\u7b97\u6cd5\u5728\u795e\u7ecf\u7f51\u7edc\uff0c\u6c42\u89e3\u504f\u5fae\u5206\u65b9\u7a0b\u95ee\u9898\u4e0a\u7684\u6027\u80fd\u5bf9\u6bd4<\/span><\/span><\/p>\n<\/section>\n

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


<\/p>\n

\n
\n
\n
\n

AI+Science \u8bfb\u4e66\u4f1a<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n


<\/span><\/section>\n
AI+Science \u662f\u8fd1\u5e74\u5174\u8d77\u7684\u5c06\u4eba\u5de5\u667a\u80fd\u548c\u79d1\u5b66\u76f8\u7ed3\u5408\u7684\u4e00\u79cd\u8d8b\u52bf\u3002<\/span>\u4e00\u65b9\u9762\u662f AI for Science\uff0c\u673a\u5668\u5b66\u4e60\u548c\u5176\u4ed6 AI \u6280\u672f\u53ef\u4ee5\u7528\u6765\u89e3\u51b3\u79d1\u5b66\u7814\u7a76\u4e2d\u7684\u95ee\u9898\uff0c\u4ece\u9884\u6d4b\u5929\u6c14\u548c\u86cb\u767d\u8d28\u7ed3\u6784\uff0c\u5230\u6a21\u62df\u661f\u7cfb\u78b0\u649e\u3001\u8bbe\u8ba1\u4f18\u5316\u6838\u805a\u53d8\u53cd\u5e94\u5806\uff0c\u751a\u81f3\u50cf\u79d1\u5b66\u5bb6\u4e00\u6837\u8fdb\u884c\u79d1\u5b66\u53d1\u73b0\uff0c\u88ab\u79f0\u4e3a\u79d1\u5b66\u53d1\u73b0\u7684\u201c\u7b2c\u4e94\u8303\u5f0f\u201d\u3002<\/span>\u53e6\u4e00\u65b9\u9762\u662f Science for AI\uff0c\u79d1\u5b66\u5c24\u5176\u662f\u7269\u7406\u5b66\u4e2d\u7684\u89c4\u5f8b\u548c\u601d\u60f3\u542f\u53d1\u673a\u5668\u5b66\u4e60\u7406\u8bba\uff0c\u4e3a\u4eba\u5de5\u667a\u80fd\u7684\u53d1\u5c55\u63d0\u4f9b\u5168\u65b0\u7684\u89c6\u89d2\u548c\u65b9\u6cd5\u3002<\/span><\/section>\n
<\/span><\/section>\n
\u96c6\u667a\u4ff1\u4e50\u90e8\u8054\u5408\u65af\u5766\u798f\u5927\u5b66\u8ba1\u7b97\u673a\u79d1\u5b66\u7cfb\u535a\u58eb\u540e\u7814\u7a76\u5458\u5434\u6cf0\u9716\uff08Jure Leskovec \u6559\u6388\u6307\u5bfc\uff09\u3001\u54c8\u4f5b\u91cf\u5b50\u8ba1\u5212\u7814\u7a76\u5458\u6248\u9e3f\u4e1a\u3001\u9ebb\u7701\u7406\u5de5\u5b66\u9662\u7269\u7406\u7cfb\u535a\u58eb\u751f\u5218\u5b50\u9e23\uff08Max Tegmark \u6559\u6388\u6307\u5bfc\uff09\uff0c\u5171\u540c\u53d1\u8d77\u4ee5\u201cAI+Science<\/a>\u201d\u4e3a\u4e3b\u9898\u7684\u8bfb\u4e66\u4f1a\uff0c\u63a2\u8ba8\u8be5\u9886\u57df\u7684\u91cd\u8981\u95ee\u9898\uff0c\u5171\u5b66\u5171\u7814\u76f8\u5173\u6587\u732e\u3002<\/span><\/span>\u8bfb\u4e66\u4f1a\u5df2\u5b8c\u7ed3\uff0c\u73b0\u5728\u62a5\u540d\u53ef\u52a0\u5165\u793e\u7fa4\u5e76\u89e3\u9501<\/span>\u56de\u653e<\/span>\u89c6\u9891\u6743\u9650\u3002<\/span><\/span><\/section>\n
<\/span><\/section>\n

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

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

<\/span><\/section>\n
\u8be6\u60c5\u8bf7\u89c1\uff1a<\/span><\/span><\/section>\n
\u4eba\u5de5\u667a\u80fd\u548c\u79d1\u5b66\u53d1\u73b0\u76f8\u4e92\u8d4b\u80fd\u7684\u65b0\u8303\u5f0f\uff1aAI+Science \u8bfb\u4e66\u4f1a\u542f\u52a8<\/a><\/span><\/span><\/section>\n

<\/section>\n

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

\u63a7\u5236\u79d1\u5b66\u524d\u6cbf\u7406\u8bba\u4e0e\u65b9\u6cd5\u7cfb\u5217\u8bfe\u7a0b<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n


<\/section>\n
\n
\u5bf9\u590d\u6742\u52a8\u6001\u7cfb\u7edf\u7684\u5b9a\u91cf\u8ba4\u8bc6\u4e0e\u79d1\u5b66\u8c03\u63a7\uff0c\u7cfb\u7edf\u5b66\u4e0e\u63a7\u5236\u8bba\u662f\u5173\u952e\u57fa\u7840\u3002\u63a7\u5236\u7406\u8bba\u662f\u4e00\u95e8\u9ad8\u5ea6\u7efc\u5408\u3001\u5145\u6ee1\u6d3b\u529b\u7684\u4ea4\u53c9\u5b66\u79d1\uff0c\u5904\u4e8e\u6570\u5b66\u3001\u8ba1\u7b97\u673a\u79d1\u5b66\u548c\u5de5\u7a0b\u6280\u672f\u79d1\u5b66\u76f8\u4e92\u4f5c\u7528\u7684\u524d\u6cbf\uff0c\u662f\u4fe1\u606f\u79d1\u5b66\u4e0e\u7cfb\u7edf\u79d1\u5b66\u7684\u91cd\u8981\u7ec4\u6210\u90e8\u5206\uff0c\u540c\u65f6\u4e0e\u81ea\u7136\u79d1\u5b66\u548c\u7ba1\u7406\u79d1\u5b66\u7684\u5404\u95e8\u5206\u652f\u4e5f\u6709\u5bc6\u5207\u7684\u8054\u7cfb\u3002\u5728\u5de5\u7a0b\u3001\u56fd\u9632\u519b\u4e8b\u3001\u793e\u4f1a\u3001\u7ecf\u6d4e\u3001\u751f\u7269\u3001\u73af\u5883\u7b49\u9886\u57df\u590d\u6742\u7cfb\u7edf\u4e2d\u7684\u8c03\u63a7\u95ee\u9898\uff0c\u53d1\u6325\u7740\u91cd\u8981\u4f5c\u7528\u3002\u300a\u63a7\u5236\u79d1\u5b66\u524d\u6cbf\u7406\u8bba\u4e0e\u65b9\u6cd5\u300b\u7cfb\u5217\u8bfe\u7a0b\uff0c\u4e3a\u9762\u5411\u63a7\u5236\u4e13\u4e1a\u9886\u57df\u5b66\u751f\u4ee5\u53ca\u8de8\u9886\u57df\u7684\u5b66\u4e60\u8005\u505a\u6df1\u5ea6\u79d1\u666e\uff0c\u5e2e\u52a9\u5b66\u4e60\u63a7\u5236\u8bba\u7684\u6838\u5fc3\u601d\u60f3\u3001\u6846\u67b6\u4e0e\u65b9\u6cd5\u3002\u8be6\u60c5\u8bf7\u89c1\uff1a<\/span>2024\u5f00\u5e74\u91cd\u78c5\u8bfe\u7a0b\uff1a\u63a7\u5236\u79d1\u5b66\u524d\u6cbf\u7406\u8bba\u4e0e\u65b9\u6cd5<\/span><\/a><\/section>\n

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

<\/section>\n

<\/section>\n<\/section>\n
\u63a8\u8350\u9605\u8bfb<\/span><\/strong><\/strong><\/strong><\/section>\n
1. <\/strong><\/a>\u90ed\u96f7\u9662\u58eb\uff1a\u7cfb\u7edf\u63a7\u5236\u4e4b\u7f8e\uff5c\u300a\u63a7\u5236\u7406\u8bba\u524d\u6cbf\u7406\u8bba\u4e0e\u65b9\u6cd5\u300b\u7b2c\u4e00\u8bfe<\/strong><\/a><\/strong><\/span><\/strong><\/strong><\/section>\n
2. <\/strong>\u65b0\u5e74\u968f\u7b14\uff1a\u4ece\u300a\u4f60\u4e00\u751f\u7684\u6545\u4e8b\u300b\u5230\u63a7\u5236\u8bba<\/a><\/strong><\/strong><\/span><\/section>\n
3. <\/strong>\u9ec4\u4e00\uff1a\u8d70\u9a6c\u89c2\u82b1\u770b\u63a7\u5236\u53d1\u5c55\u7b80\u53f2<\/a><\/strong><\/span><\/strong><\/strong><\/section>\n
4. <\/strong><\/span>\u5f20\u6c5f\uff1a\u7b2c\u4e09\u4ee3\u4eba\u5de5\u667a\u80fd\u6280\u672f\u57fa\u7840\u2014\u2014\u4ece\u53ef\u5fae\u5206\u7f16\u7a0b\u5230\u56e0\u679c\u63a8\u7406 | \u96c6\u667a\u5b66\u56ed\u5168\u65b0\u8bfe\u7a0b<\/a><\/strong><\/strong><\/span><\/section>\n
<\/strong>5. <\/strong>\u9f99\u5e74\u5927\u8fd0\u8d77\uff0c\u5b66\u4e60\u6b63\u5f53\u65f6\uff01\u89e3\u9501\u96c6\u667a\u5168\u7ad9\u5185\u5bb9\uff0c\u5f00\u542f\u65b0\u5e74\u5b66\u4e60\u8ba1\u5212<\/a><\/strong><\/section>\n

6. <\/strong>\u52a0\u5165\u96c6\u667a\uff0c\u4e00\u8d77\u590d\u6742\uff01<\/strong><\/span><\/a><\/p>\n


<\/p>\n


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

\u70b9\u51fb\u201c\u9605\u8bfb\u539f\u6587\u201d\uff0c\u62a5\u540d\u8bfb\u4e66\u4f1a<\/span><\/section>\n

<\/mp-style-type><\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

\u5bfc\u8bed \u8fd1\u65e5\uff0cNature Communications \u53d1\u8868\u9898\u4e3a\u201cAccelerated optimization in deep learning with a proportional-integral derivative controller\u201d\u7684\u6700\u65b0\u8bba\u6587\uff0c\u8be5\u7814\u7a76\u5c06\u95ed\u73af\u53cd\u9988\u63a7\u5236\u4e0e\u4f18\u5316\u7b97\u6cd5\u5efa\u7acb\u8054…<\/p>\n","protected":false},"author":0,"featured_media":54084,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[1],"tags":[],"special":[],"_links":{"self":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/54091"}],"collection":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=54091"}],"version-history":[{"count":0,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/54091\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/media\/54084"}],"wp:attachment":[{"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=54091"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=54091"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=54091"},{"taxonomy":"special","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fspecial&post=54091"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}