{"id":21502,"date":"2020-10-28T22:05:19","date_gmt":"2020-10-28T14:05:19","guid":{"rendered":"https:\/\/swarma.org\/?p=21502"},"modified":"2020-10-28T22:05:19","modified_gmt":"2020-10-28T14:05:19","slug":"%e5%a4%8d%e6%9d%82%e7%b3%bb%e7%bb%9f%e8%87%aa%e5%8a%a8%e5%bb%ba%e6%a8%a1%e7%bb%bc%e8%bf%b0%ef%bc%9a%e6%8f%8f%e8%bf%b0%e3%80%81%e9%a2%84%e6%b5%8b%e4%b8%8e%e7%90%86%e8%ae%ba%e5%8f%91%e7%8e%b0","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=21502","title":{"rendered":"\u590d\u6742\u7cfb\u7edf\u81ea\u52a8\u5efa\u6a21\u7efc\u8ff0\uff1a\u63cf\u8ff0\u3001\u9884\u6d4b\u4e0e\u7406\u8bba\u53d1\u73b0"},"content":{"rendered":"
\n

<\/p>\n


<\/p>\n

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

\u590d\u6742\u7cfb\u7edf\u5728\u6211\u4eec\u65e5\u5e38\u751f\u6d3b\u4e2d\u65e0\u5904\u4e0d\u5728\uff0c\u4f46\u65e0\u8bba\u662f\u5bf9\u590d\u6742\u7cfb\u7edf\u7684\u523b\u753b\u63cf\u8ff0\uff0c\u8fd8\u662f\u5bf9\u590d\u6742\u7cfb\u7edf\u8fdb\u884c\u6709\u6548\u7684\u9884\u6d4b\uff0c\u4ee5\u53ca\u79d1\u5b66\u7406\u8bba\u53d1\u73b0\uff0c\u90fd\u662f\u4e00\u9879\u5de8\u5927\u7684\u6311\u6218\u3002\u672c\u6587\u603b\u7ed3\u4e86\u5f20\u6c5f\u8001\u5e08\u5728\u96c6\u667a-\u51ef\u98ce\u8bfb\u4e66\u4f1a\u4e0a\u7684\u300a\u590d\u6742\u7cfb\u7edf\u81ea\u52a8\u5efa\u6a21\u300b\u4e13\u9898\uff0c\u4ecb\u7ecd\u4e86\u5982\u4f55\u66f4\u597d\u5730\u5229\u7528AI\u6765\u5bf9\u590d\u6742\u7cfb\u7edf\u66f4\u597d\u7684\u8fdb\u884c\u63cf\u8ff0\u3001\u9884\u6d4b\u4e43\u81f3\u4e8e\u7406\u8bba\u53d1\u73b0\uff0c\u6700\u540e\uff0c\u672c\u6587\u8fd8\u5bf9\u901a\u8fc7AI\u5bf9\u590d\u6742\u7cfb\u7edf\u8fdb\u884c\u81ea\u52a8\u5efa\u6a21\u4e2d\u7684\u4e00\u4e9b\u53ef\u80fd\u7684\u95ee\u9898\u8fdb\u884c\u4e86\u7b80\u5355\u7684\u63cf\u8ff0\u548c\u8ba8\u8bba\u3002<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

 <\/span>
<\/p>\n

<\/p>\n

\u626b\u7801\u67e5\u770b\u590d\u6742\u7cfb\u7edf\u81ea\u52a8\u5efa\u6a21\u7684\u8bba\u6587\u8d44\u6599\u6574\u7406<\/span><\/p>\n

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


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

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

\u4e3a\u4ec0\u4e48\u8981\u5229\u7528AI\u63a2\u7d22\u590d\u6742\u7cfb\u7edf<\/strong><\/span><\/p>\n

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

 <\/span><\/p>\n

\n
\u4f20\u7edf\u6a21\u5f0f\u4e0b\u5bf9\u590d\u6742\u7cfb\u7edf\u8fdb\u884c\u5efa\u6a21<\/span><\/strong><\/span><\/section>\n<\/h3>\n
 <\/span><\/section>\n
\u8bb8\u591a\u76f8\u4e92\u4f5c\u7528\u7684\u5143\u7d20\u53ef\u4ee5\u7ec4\u6210\u590d\u6742\u7cfb\u7edf\uff0c\u5982\u4e92\u8054\u7f51\u3001\u7ecf\u6d4e\u7cfb\u7edf\u3001\u516c\u5171\u90e8\u95e8\u548c\u884c\u653f\u7ec4\u7ec7\u3001\u57ce\u5e02\u3001\u751f\u547d\u4f53\u3001\u5927\u8111\u3001\u7535\u7f51\u3001\u4ea4\u901a\u3001\u751f\u6001\u7cfb\u7edf\u7b49\u7b49\uff0c\u590d\u6742\u7cfb\u7edf\u5b58\u5728\u4e8e\u6211\u4eec\u751f\u6d3b\u7684\u5404\u4e2a\u9886\u57df\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u5404\u5f0f\u590d\u6742\u7cfb\u7edf\u5f80\u5f80\u53ef\u4ee5\u88ab\u62bd\u8c61\u4e3a\u4e00\u7ec4\u5355\u5143\uff0c\u5b83\u4eec\u901a\u8fc7\u4e00\u4e2a\u7f51\u7edc\u76f8\u8fde\uff0c\u5e76\u6309\u7167\u4e00\u5b9a\u7684\u52a8\u529b\u5b66\u6cd5\u5219\u53d1\u751f\u76f8\u4e92\u4f5c\u7528\u3002\u7136\u800c\uff0c\u7531\u4e8e\u590d\u6742\u7cfb\u7edf\u4e2d\u7684\u5143\u7d20\u3001\u4e3b\u4f53\u6570\u91cf\u8f83\u591a\uff0c\u7cfb\u7edf\u7ec4\u6210\u5143\u7d20\u4e4b\u95f4\u4ee5\u53ca\u7cfb\u7edf\u548c\u73af\u5883\u4e4b\u95f4\u5b58\u5728\u4f9d\u8d56\u3001\u7ade\u4e89\u3001\u5173\u8054\u7b49\u590d\u6742\u7684\u4f5c\u7528\uff0c\u4fbf\u4ea7\u751f\u4e86\u5982\u975e\u7ebf\u6027 \u3001\u6d8c\u73b0\u3001\u81ea\u53d1\u79e9\u5e8f\u3001\u9002\u5e94\u6027\u4ee5\u53ca\u53cd\u9988\u56de\u8def\u7b49\u7684\u7279\u6b8a\u6027\u8d28\uff0c\u5bfc\u81f4\u4e92\u52a8\u6a21\u5f0f\u4e0e\u5f71\u54cd\u96be\u4ee5\u63cf\u8ff0\uff0c\u53c2\u6570\u6570\u91cf\u7684\u589e\u52a0\u6709\u53ef\u80fd\u5feb\u4e8e\u7cfb\u7edf\u5927\u5c0f\u7684\u589e\u52a0\uff0c\u6211\u4eec\u65e2\u65e0\u6cd5\u628a\u6574\u4e2a\u7cfb\u7edf\u57fa\u4e8e\u8fd8\u539f\u8bba\u800c\u7b80\u5355\u7406\u89e3\u4e3a\u4e2a\u4f53\u7684\u52a0\u603b\uff0c\u4e5f\u65e0\u6cd5\u4ece\u6574\u4f53\u7684\u884c\u4e3a\u6709\u6548\u63a8\u65ad\u51fa\u7cfb\u7edf\u7684\u786e\u5b9a\u6027\u8d28\u3002<\/span><\/section>\n
 <\/span><\/section>\n
\u6240\u4ee5\uff0c\u6211\u4eec\u5bf9\u590d\u6742\u7cfb\u7edf\u7684\u5efa\u6a21\u662f\u6781\u5176\u56f0\u96be\u7684\u3002<\/span><\/section>\n
 <\/span> <\/span><\/section>\n

19\u4e16\u7eaa\u300120\u4e16\u7eaa\u7279\u522b\u662f20\u4e16\u7eaa80\u5e74\u4ee3\u4ee5\u6765\uff0c\u590d\u6742\u79d1\u5b66\u3001\u7f51\u7edc\u79d1\u5b66\u7684\u4e0d\u65ad\u53d1\u5c55\u4f7f\u5f97\u6211\u4eec\u53ef\u4ee5\u4ece\u66f4\u591a\u89d2\u5ea6\u53bb\u8ba4\u8bc6\u3001\u7814\u7a76\u3001\u523b\u753b\u548c\u9884\u6d4b\u590d\u6742\u73b0\u8c61\u3002\u6bd4\u5982\u5bf9\u4e8e\u57ce\u5e02\u8fd9\u4e00\u590d\u6742\u7cfb\u7edf\u7684\u6f14\u5316\u589e\u957f\uff0c\u867d\u7136\u6211\u4eec\u4e0d\u671f\u5f85\u7528\u4e00\u5f20\u9910\u5dfe\u7eb8\u5199\u4e0b\u57ce\u5e02\u7684\u751f\u957f\u65b9\u7a0b\uff0c\u4f46\u6211\u4eec\u7edd\u5bf9\u53ef\u4ee5\u7528\u4e00\u9875\u7eb8\u5199\u4e0b\u6240\u6709\u7684\u89c4\u5219\uff1a<\/span><\/p>\n

    \n
  • \n

    \u7a7a\u95f4\u5438\u5f15\u89c4\u5219\uff1a\u57ce\u5e02\u4e2d\u67d0\u4e00\u5730\u70b9\u5bf9\u65b0\u52a0\u5165\u4e2a\u4f53<\/span>\uff08\u4ee5\u4e8c\u7ef4\u5e73\u9762\u4e0a\u7684\u8282\u70b9\u4e3a\u4ee3\u8868\uff09<\/span>\u7684\u5438\u5f15\u529b\u6b63\u6bd4\u4e8e\u5176\u81ea\u7136\u7980\u8d4b\u4e0e\u793e\u4f1a\u5438\u5f15<\/span>\uff08\u5728\u6a21\u578b\u4e2d\u6211\u4eec\u7528\u5f53\u5730\u7684\u6d3b\u8dc3\u4eba\u53e3\u5bc6\u5ea6\u6765\u8868\u793a\uff09<\/span>\u4e4b\u548c\uff1b<\/span><\/p>\n<\/li>\n

  • \n

    \u5339\u914d\u751f\u957f\u89c4\u5219\uff1a\u5982\u679c\u65b0\u52a0\u5165\u7cfb\u7edf\u7684\u8282\u70b9<\/span>\uff08\u4ee3\u8868\u6d3b\u8dc3\u793e\u533a\uff09<\/span>\u79bb\u73b0\u6709\u57ce\u5e02\u4eba\u53e3\u8fc7\u8fdc\uff0c\u90a3\u4e48\u5b83\u5c31\u65e0\u6cd5\u6301\u7eed\u5b58\u5728<\/span>\uff08\u89c1\u4e0b\u56fea\uff09<\/span>\uff1b<\/span><\/p>\n<\/li>\n

  • \n

    \u9053\u8def\u751f\u957f\u89c4\u5219\uff1a\u9053\u8def\u7f51\u7edc\u7684\u5bc6\u5ea6\u7531\u4eba\u53e3\u5bc6\u5ea6\u51b3\u5b9a\u3002\u6700\u7b80\u5355\u7684\u8def\u7f51\u6784\u5efa\u539f\u5219\u5c31\u662f\u4f9d\u6700\u7b80\u5355\u7684\u6cf0\u68ee\u591a\u8fb9\u5f62\u5212\u5206\u8fdb\u884c\u8def\u7f51\u6784\u5efa<\/span>\uff08\u5982\u4e0b\u56fec\uff09<\/span>\uff1b<\/span><\/p>\n<\/li>\n

  • \n

    \u793e\u4f1a\u4e92\u52a8\u89c4\u5219\uff1a\u5e02\u6c11\u7684\u4e92\u52a8\u662f\u6cbf\u7740\u57ce\u5e02\u7684\u754c\u9762\u2014\u2014\u8857\u9053\u5c55\u5f00\u7684\uff0c\u56e0\u6b64\u4e92\u52a8\u7684\u5bc6\u5ea6\u7b49\u4e8e\u4eba\u53e3\u5bc6\u5ea6\u548c\u8def\u7f51\u5bc6\u5ea6\u7684\u4e58\u79ef\u3002\u540c\u65f6\u6211\u4eec\u5047\u8bbe\u8fd9\u4e9b\u6d3b\u52a8\u521b\u9020\u4e86\u57ce\u5e02\u7684\u7ecf\u6d4e\u4ef7\u503c\u548c\u521b\u65b0<\/span>\uff08\u89c1\u4e0b\u56feb\uff09<\/span>\u3002<\/span> <\/span>             <\/span><\/p>\n<\/li>\n<\/ul>\n


    <\/section>\n
    <\/span><\/section>\n
    \u56fe1\uff1a\u590d\u6742\u7684\u57ce\u5e02\uff0c\u7b80\u5355\u7684\u89c4\u5219<\/span><\/section>\n

    <\/span><\/section>\n
    \n
    Li, R., Dong, L., Zhang, J., Wang, X., Wang, W. X., & Di, Z & Stanley, H. E. (2017). Simple spatial scaling rules behind complex cities. Nature Communications, 8(1), 1841.     \u4e0b\u8f7d\u5730\u5740\uff1ahttps:\/\/www.nature.com\/articles\/s41467-017-01882-w\u3002<\/span><\/section>\n<\/blockquote>\n
     <\/span><\/section>\n
    \u6211\u4eec\u7684\u53e6\u4e00\u9879\u5de5\u4f5c\u662f\u5bf9\u65b0\u51a0\u75ab\u60c5\u5728\u57ce\u5e02\u95f4\u7684\u4f20\u64ad\u9632\u63a7\u8fdb\u884c\u5efa\u6a21\u3002\u75ab\u60c5\u671f\u95f4\u6211\u4eec\u6210\u7acb\u4e86\u96c6\u667a\u75ab\u60c5\u5c0f\u7ec4\uff0c\u5c1d\u8bd5\u901a\u8fc7\u5bf9\u75ab\u60c5\u4f20\u64ad\u8fdb\u884c\u5efa\u6a21\u5206\u6790\u4ece\u800c\u5f97\u5230\u4e00\u4e9b\u53ef\u4ee5\u5e2e\u52a9\u9632\u63a7\u75ab\u60c5\u7684\u77e5\u8bc6\u3002\u5bf9\u4e8e\u75be\u75c5\u4f20\u64ad\uff0c\u6211\u4eec\u5c06\u57ce\u5e02\u4f5c\u4e3a\u8282\u70b9\uff0c\u57ce\u5e02\u611f\u67d3\u4eba\u6570\u4f5c\u4e3a\u8282\u70b9\u7684\u7279\u5f81\uff0c\u6839\u636e\u57ce\u5e02\u95f4\u7684\u4eba\u53e3\u8fc1\u79fb\u6570\u636e\u5efa\u7acb\u6574\u4e2a\u7f51\u7edc\uff0c\u901a\u8fc7\u6570\u636e\u9a71\u52a8\u7684\u65b9\u5f0f\u5bf9Meta-population model\u4e2d\u7684\u53c2\u6570\u8fdb\u884c\u5b66\u4e60\u3002\u867d\u7136\u6211\u4eec\u8fd9\u9879\u7814\u7a76\u8fd0\u7528\u4e86AI\uff0c\u4f46\u6211\u4eec\u53ea\u662f\u5229\u7528\u5176\u5b66\u4e60\u6a21\u578b\u4e2d\u7684\u53c2\u6570\uff0c\u6a21\u578b\u672c\u8eab\u7684\u6784\u5efa\u4ecd\u65e7\u4f9d\u8d56\u79d1\u5b66\u5bb6\u5bf9\u590d\u6742\u7cfb\u7edf\u7684\u6d1e\u89c1\u3002<\/span> <\/span><\/section>\n
           <\/span>   <\/span>\u56fe2\uff1a\u65b0\u51a0\u75ab\u60c5\u5728\u57ce\u5e02\u95f4\u7684\u4f20\u64ad<\/span><\/section>\n

    <\/span><\/section>\n
    \n
    Jiang Zhang, Lei Dong, Yanbo Zhang.et al.. Investigating time, strength, and duration of measures in controlling the spread of COVID-19 using a networked meta-population model. Nonlinear Dynamics, 2020<\/span><\/section>\n<\/blockquote>\n
     <\/span><\/section>\n
    \u53ef\u4ee5\u770b\u5230\uff0c\u590d\u6742\u79d1\u5b66\u548c\u7f51\u7edc\u79d1\u5b66\u7684\u4e0d\u65ad\u53d1\u5c55\u4f7f\u5f97\u6211\u4eec\u6709\u751f\u547d\u6e38\u620f\u3001Boid\u6a21\u578b\u3001SIR\u75c5\u6bd2\u4f20\u64ad\u6a21\u578b\u3001\u504f\u597d\u4f9d\u9644\u7f51\u7edc\u589e\u957f\u6a21\u578b\u3001\u5339\u914d\u751f\u957f\u6a21\u578b\u3001\u4eba\u5de5\u80a1\u5e02\u6a21\u578b\u7b49\u7b49\u53ef\u4ee5\u7528\u6765\u5bf9\u4e00\u4e9b\u590d\u6742\u7cfb\u7edf\u8fdb\u884c\u8f83\u597d\u7684\u63cf\u8ff0\u3001\u523b\u753b\u548c\u884c\u4e3a\u9884\u6d4b\uff0c\u8fd9\u4e9b\u7b80\u6d01\u3001\u5bcc\u6709\u6d1e\u5bdf\u529b\u3001\u4fbf\u4e8e\u7406\u8bba\u53d1\u73b0\u5206\u6790\u4e14\u8ba1\u7b97\u5ec9\u4ef7\u7684\u6a21\u578b\u5927\u5927\u52a0\u6df1\u4e86\u6211\u4eec\u5bf9\u590d\u6742\u7cfb\u7edf\u7684\u8ba4\u8bc6\uff1b\u7136\u800c\u4f9d\u9760\u4eba\u5de5\u6784\u5efa\u6a21\u578b\u9762\u4e34\u7740\u5f88\u591a\u7f3a\u70b9\uff1a<\/span><\/section>\n
     <\/span><\/section>\n
      \n
    • \n

      \u8fc7\u4e8e\u7b80\u5355\uff0c\u4ee5\u81f3\u4e8e\u5f88\u96be\u89e3\u91ca\u66f4\u591a\u7684\u590d\u6742\u73b0\u8c61<\/span><\/p>\n<\/li>\n

    • \n

      \u5f88\u96be\u4e0e\u89c2\u6d4b\u6570\u636e\u878d\u5408<\/span><\/p>\n<\/li>\n

    • \n

      \u9884\u6d4b\u7cbe\u5ea6\u8f83\u4f4e<\/span><\/p>\n<\/li>\n

    • \n

      \u4f9d\u8d56\u4e8e\u4eba\u7c7b\u5efa\u6a21\u8005\u7684\u542f\u53d1\u6027\u601d\u8003<\/span><\/p>\n<\/li>\n

    • \n

      \u5efa\u6a21\u6ca1\u6709\u7edf\u4e00\u7684\u89c4\u5219<\/span><\/p>\n<\/li>\n<\/ul>\n

       <\/span><\/section>\n

      \n
      \u5229\u7528AI\u5bf9\u590d\u6742\u7cfb\u7edf\u8fdb\u884c\u81ea\u52a8\u5efa\u6a21<\/span><\/strong><\/span><\/section>\n<\/h3>\n
       <\/span><\/section>\n
      2007\u5e741\u670828\u65e5\uff0c\u56fe\u7075\u5956\u5f97\u4e3b\u5409\u59c6\u00b7\u683c\u96f7<\/span> (Jim Gray) <\/span>\u5728\u81ea\u5df1\u9177\u7231\u7684\u822a\u6d77\u8fd0\u52a8\u4e2d\u9a7e\u9a76\u5e06\u8239\u5931\u8e2a\u4e8e\u832b\u832b\u5927\u6d77\u4e4b\u4e0a\u3002\u77ed\u77ed17\u5929\u4e4b\u524d\uff0c\u4ed6\u5728\u52a0\u5dde\u5c71\u666f\u57ce\u53ec\u5f00\u7684NRC-CSTB<\/span>\uff08National Research Council-Computer Science and Telecommunications Board\uff09<\/span>\u4e0a\uff0c\u53d1\u8868\u4e86\u4ed6\u7684\u8457\u540d\u6f14\u8bb2\uff1aThe Fourth Paradigm: Data-Intensive Scientific Discovery<\/span> (\u300a\u7b2c\u56db\u8303\u5f0f\uff1a\u6570\u636e\u5bc6\u96c6\u578b\u79d1\u5b66\u53d1\u73b0\u300b)<\/span>\u3002\u6f14\u8bb2\u4e2d\uff0c\u4ed6\u5c06\u4eba\u7c7b\u79d1\u5b66\u53d1\u5c55\u5206\u4e3a\u56db\u79cd\u8303\u5f0f\uff1a\u4ee5\u8bb0\u5f55\u548c\u63cf\u8ff0\u81ea\u7136\u73b0\u8c61\u4e3a\u4e3b\u7684\u201c\u5b9e\u9a8c\u79d1\u5b66\u201d\uff0c\u5373\u7b2c\u4e00\u8303\u5f0f\uff1b\u5229\u7528\u6a21\u578b\u5f52\u7eb3\u603b\u7ed3\u8fc7\u53bb\u8bb0\u5f55\u7684\u73b0\u8c61\u7684\u201c\u7406\u8bba\u79d1\u5b66\u201d\uff0c\u5373\u7b2c\u4e8c\u8303\u5f0f\uff1b\u79d1\u5b66\u8ba1\u7b97\u673a\u7684\u51fa\u73b0\u5e26\u6765\u7684\u6a21\u62df\u4eff\u771f\u7b49\u201c\u8ba1\u7b97\u79d1\u5b66\u201d\uff0c\u5373\u7b2c\u4e09\u8303\u5f0f\uff1b\u5982\u4eca\u5927\u6570\u636e\u65f6\u4ee3\u7684\u5230\u6765\uff0c\u51fa\u73b0\u65b0\u7684\u79d1\u5b66\u7814\u7a76\u65b9\u5f0f\uff0c\u5373\u6570\u636e\u5bc6\u96c6\u578b\u79d1\u5b66<\/span>\uff08Data-Intensive Scientific Discovery\uff09<\/span>\uff0c\u79f0\u4e3a\u7b2c\u56db\u8303\u5f0f\u3002<\/span><\/section>\n

      <\/section>\n
      <\/span><\/section>\n
      \u56fe3\uff1a\u79d1\u5b66\u7684\u56db\u79cd\u7814\u7a76\u8303\u5f0f<\/span><\/section>\n

      <\/span><\/section>\n
      \n
      \u7c73\u52a0\u5b81,\u7ae0\u660c\u5e73,\u674e\u5927\u5b87 & \u6797\u6d9b.(2018).\u7b2c\u56db\u7814\u7a76\u8303\u5f0f:\u5927\u6570\u636e\u9a71\u52a8\u7684\u793e\u4f1a\u79d1\u5b66\u7814\u7a76\u8f6c\u578b. \u5b66\u6d77(02),11-27. doi:10.16091\/j.cnki.cn32-1308\/c.2018.02.003.<\/span><\/section>\n<\/blockquote>\n
       <\/span><\/section>\n
      \u800c\u7b2c\u56db\u8303\u5f0f\u4e0e\u7b2c\u4e09\u8303\u5f0f\u6700\u663e\u8457\u7684\u533a\u522b\uff0c\u5728\u4e8e\u7b2c\u4e09\u8303\u5f0f\u662f\u5148\u63d0\u51fa\u53ef\u80fd\u7684\u7406\u8bba\uff0c\u518d\u641c\u96c6\u6570\u636e\uff0c\u7136\u540e\u901a\u8fc7\u8ba1\u7b97\u4eff\u771f\u8fdb\u884c\u7406\u8bba\u9a8c\u8bc1\u3002\u800c\u7b2c\u56db\u8303\u5f0f\u662f\u5148\u6709\u4e86\u5927\u91cf\u7684\u5df2\u77e5\u6570\u636e\uff0c\u7136\u540e\u901a\u8fc7\u8ba1\u7b97\u5f97\u51fa\u4e4b\u524d\u672a\u77e5\u7684\u53ef\u4fe1\u7684\u7406\u8bba\u3002\u663e\u7136\uff0c\u7b2c\u56db\u8303\u5f0f\u7684\u6838\u5fc3\uff0c\u5728\u4e8e\u4e00\u79cd\u65b0\u7684\u8ba4\u77e5\u9014\u5f84\uff1a\u5bfb\u627e\u201c\u4ea7\u751f\u4e8e\u6570\u636e\u201d<\/span>\uff08born from the data\uff09<\/span>\u4e2d\u7684\u6d1e\u89c1\u3002\u800c\u5927\u6570\u636e\u7684\u7279\u70b9\u662f\u8fde\u7eed\u4ea7\u751f\u7684\u6570\u636e\uff0c\u5e76\u529b\u6c42\u5728\u8303\u56f4\u4e0a\u7a77\u4e3e\u548c\u5728\u5185\u5bb9\u4e0a\u7cbe\u7ec6\uff0c\u540c\u65f6\u6570\u636e\u4ea7\u51fa\u4e0a\u53c8\u662f\u5f39\u6027\u7075\u6d3b\u3001\u5927\u5c0f\u53ef\u53d8\u7684\u3002\u6211\u4eec\u8ba4\u8bc6\u5230\uff0c\u968f\u7740\u5927\u6570\u636e\u7684\u79ef\u7d2f\u548c\u4eba\u5de5\u667a\u80fd\u7684\u4e0d\u65ad\u53d1\u5c55\uff0c\u7279\u522b\u662f\u6df1\u5ea6\u5b66\u4e60\u6280\u672f\u7684\u53d1\u5c55\uff0c\u4f7f\u5f97\u6211\u4eec\u53ef\u4ee5\u901a\u8fc7\u6570\u636e\u9a71\u52a8<\/span>\uff08data-driven science\uff09<\/span>\u7684\u65b9\u5f0f\uff0c\u5229\u7528\u6df1\u5ea6\u5b66\u4e60\u7b97\u6cd5\u6765\u81ea\u52a8\u6784\u5efa\u590d\u6742\u7cfb\u7edf\u7684\u6a21\u578b\uff0c\u6211\u4eec\u62e5\u6709\u4e86\u66f4\u52a0\u5f3a\u529b\u7684\u5de5\u5177\u6765\u5bf9\u590d\u6742\u7cfb\u7edf\u8fdb\u884c\u66f4\u52a0\u7cbe\u786e\u3001\u6df1\u5165\u7684\u5206\u6790\u3001\u6a21\u62df\u3001\u9884\u6d4b\uff0c\u751a\u81f3\u79d1\u5b66\u53d1\u73b0\u3002<\/span><\/section>\n

      <\/span><\/section>\n
      \n
      \u53ef\u53c2\u8003\uff1a<\/span>Rob Kitchin: Big Data, New Epistemologies and Paradigm Shifts, Big Data & Society, vol. 1 no. 1, 2014, pp. 1\u201312.<\/span><\/section>\n<\/blockquote>\n
       <\/span><\/section>\n
      \u663e\u7136\uff0c\u5229\u7528AI\u81ea\u52a8\u5efa\u6a21\u8f83\u4f20\u7edf\u4eba\u5de5\u5efa\u6a21<\/span>\uff08\u57fa\u4e8e\u7b2c\u4e8c\u3001\u7b2c\u4e09\u8303\u5f0f\uff09<\/span>\u6709\u663e\u8457\u7684\u533a\u522b\uff0c\u81ea\u52a8\u5efa\u6a21\u5bfb\u6c42\u201c\u4ea7\u751f\u4e8e\u6570\u636e\u201d\u800c\u4e0d\u662f\u201c\u4ea7\u751f\u4e8e\u7406\u8bba\u201d\u7684\u5047\u8bbe\u548c\u6d1e\u89c1\uff0c\u5373\u5148\u6709\u4e00\u4e2a\u4ece\u6570\u636e\u4e2d\u4ee5\u5f52\u7eb3\u7684\u65b9\u5f0f\u6765\u5f62\u6210\u67d0\u79cd\u5047\u8bbe\uff0c\u7136\u540e\u518d\u7528\u6f14\u7ece\u7684\u65b9\u5f0f\u8fdb\u884c\u7814\u7a76\u3002\u8fd9\u79cd\u6570\u636e\u83b7\u53d6\u548c\u5206\u6790\u7684\u51b3\u7b56\u601d\u8def\uff0c\u5efa\u7acb\u5728\u6eaf\u56e0\u63a8\u7406<\/span> (abductive reasoning) <\/span>\u4e4b\u4e0a\u3002\u8fd9\u79cd\u8303\u5f0f\u66f4\u52a0\u9002\u7528\u4e8e\u63d0\u70bc\u4f20\u7edf\u201c\u77e5\u8bc6\u9a71\u52a8\u7684\u79d1\u5b66\u201d\u96be\u4ee5\u63d0\u70bc\u7684\u989d\u5916\u7684\u3001\u6709\u4ef7\u503c\u7684\u6d1e\u89c1\u3002<\/span><\/section>\n
       <\/span><\/section>\n
      \u5176\u6b21\uff0c\u968f\u7740\u6280\u672f\u7684\u53d1\u5c55\uff0c\u6211\u4eec\u5df2\u7ecf\u79ef\u7d2f\u4e86\u5173\u4e8e\u590d\u6742\u7cfb\u7edf\u8fd0\u8f6c\u7684\u5927\u91cf\u6570\u636e\uff0c\u7279\u522b\u662f\u7cfb\u7edf\u6240\u4ea7\u751f\u7684\u65f6\u95f4\u5e8f\u5217\u6570\u636e\uff0c\u800c\u4f20\u7edf\u4eba\u5de5\u5efa\u6a21\u65b9\u6cd5\u5f88\u96be\u4e0e\u8fd9\u79cd\u7ed3\u6784\u7684\u6570\u636e\u76f8\u7ed3\u5408\uff0c\u6362\u8a00\u4e4b\uff0c\u4f20\u7edf\u5efa\u6a21\u65b9\u6cd5\u96be\u4ee5\u63d0\u53d6\u6570\u636e\u4e2d\u9690\u85cf\u7684\u5927\u91cf\u4fe1\u606f\u3001\u77e5\u8bc6\u3002\u4e0d\u8fc7\uff0cAI\u4e2d\u7684\u4e00\u4e9b\u7b97\u6cd5\u5374\u975e\u5e38\u64c5\u957f\u4e8e\u4ece\u65f6\u95f4\u5e8f\u5217\u6570\u636e\u7b49\u4e2d\u5b66\u4e60\u63d0\u53d6\u6a21\u578b\uff0c\u540c\u65f6\uff0c\u5f53\u524d\u7684\u786c\u4ef6\u5df2\u7ecf\u80fd\u591f\u652f\u6301\u6211\u4eec\u501f\u52a9\u4e00\u4e9b\u6df1\u5ea6\u5b66\u4e60\u7684\u7b97\u6cd5\u5bf9\u8f83\u5927\u91cf\u7ea7\u7684\u6570\u636e\u8fdb\u884c\u5b66\u4e60\u5206\u6790\u3002<\/span><\/section>\n
       <\/span><\/section>\n
      \u56e0\u6b64\uff0c\u6211\u4eec\u4fbf\u6709\u53ef\u80fd\u5728AI\u7684\u5e2e\u52a9\u4e0b\u5b8c\u6210\u5bf9\u590d\u6742\u7cfb\u7edf\u7684\u81ea\u52a8\u5efa\u6a21\uff0c\u4f7f\u5f97\u6211\u4eec\u4e0d\u4ec5\u4ec5\u80fd\u591f\u5bf9\u6a21\u578b\u53c2\u6570\u8fdb\u884c\u5b66\u4e60\u8ba1\u7b97\uff0c\u66f4\u91cd\u8981\u7684\u662f\u5229\u7528AI\u81ea\u52a8\u5bf9\u6a21\u578b\u672c\u8eab\u8fdb\u884c\u5b66\u4e60\u6784\u5efa\uff0c\u6700\u7ec8\u5b9e\u73b0\u5bf9\u590d\u6742\u7cfb\u7edf\u7684\u63cf\u8ff0\u3001\u9884\u6d4b\u548c\u79d1\u5b66\u53d1\u73b0\u3002<\/span><\/section>\n
       <\/span><\/section>\n

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

      AI\u81ea\u52a8\u5efa\u6a21\u7684\u4e94\u4e2a\u9636\u6bb5<\/strong><\/span><\/p>\n

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

       <\/span><\/p>\n

      AI\u81ea\u52a8\u5efa\u6a21\uff0c\u7b80\u5355\u6765\u8bf4\u5c31\u662f\u6839\u636e\u83b7\u53d6\u5230\u7684\u590d\u6742\u7cfb\u7edf\u7684\u89c2\u6d4b\u6570\u636e\uff0c\u5229\u7528 AI \u7cfb\u7edf\uff0c\u81ea\u52a8\u53bb\u6784\u5efa\u51fa\u4e00\u4e2a\u6211\u4eec\u60f3\u8981\u7684\u6a21\u578b\uff0c\u4ece\u800c\u6355\u83b7\u5230\u590d\u6742\u7cfb\u7edf\u5185\u90e8\u7684\u89c4\u5219\u3002\u8fd9\u6837\uff0c\u6211\u4eec\u5c31\u80fd\u591f\u5bf9\u771f\u5b9e\u7684\u590d\u6742\u7cfb\u7edf\u8fdb\u884c\u63cf\u8ff0\u523b\u753b\uff0c\u8fdb\u800c\u505a\u51fa\u5206\u6790\u9884\u6d4b\uff0c\u5b8c\u6210\u79d1\u5b66\u53d1\u73b0\uff0c\u6700\u7ec8\u5b9e\u73b0\u5bf9\u590d\u6742\u7cfb\u7edf\u7684\u8c03\u63a7\u548c\u6784\u9020\u3002<\/span><\/section>\n
                   <\/span><\/section>\n
      <\/span><\/section>\n
      \u56fe4\uff1aAI\u81ea\u52a8\u5efa\u6a21\u7684\u6d41\u7a0b<\/span><\/section>\n
       <\/span><\/section>\n
      \u6211\u4eec\u5927\u4f53\u6309\u7167\u53d1\u5c55\u987a\u5e8f\uff0c\u5c06\u8fd9\u4e00\u9886\u57df\u5206\u6210\u4e86\u5982\u4e0b\u4e94\u4e2a\u9636\u6bb5<\/span>\uff08\u65b9\u9762\uff09<\/span>\uff1a<\/span><\/section>\n

      <\/section>\n
      <\/span><\/section>\n
      \u56fe5\uff1aAI\u81ea\u52a8\u5efa\u6a21\u4e94\u90e8\u66f2<\/span><\/section>\n
       <\/span><\/section>\n
      \u6211\u4eec\u5c06\u7b80\u8981\u4ecb\u7ecd\u5404\u4e2a\u9636\u6bb5\u7684\u4ee3\u8868\u6027\u6848\u4f8b\uff0c\u540c\u65f6\u7ed3\u5408\u6211\u7684\u4e00\u4e9b\u76f8\u5173\u7814\u7a76\u8fdb\u884c\u5c55\u793a\u3002<\/span><\/section>\n
       <\/span><\/section>\n

      \n
      \u9636\u6bb51\uff1a\u5bf9\u590d\u6742\u7cfb\u7edf\u7684\u957f\u7a0b\u5173\u8054\u81ea\u52a8\u5efa\u6a21\u2014\u2014RNN<\/span><\/strong><\/span><\/section>\n<\/h3>\n
       <\/span><\/section>\n
      \u590d\u6742\u7cfb\u7edf\u7684\u884c\u4e3a\u6570\u636e\u5927\u591a\u8868\u73b0\u4e3a\u65f6\u95f4\u5e8f\u5217\uff0c\u800cRNN\u662f\u65f6\u95f4\u5e8f\u5217\u9884\u6d4b\u7684\u4e3b\u8981\u5de5\u5177\uff0c\u56e0\u6b64\u7528RNN\u6765\u81ea\u52a8\u5efa\u6a21\u662f\u4e00\u4e2a\u4e3b\u529b\u65b9\u5411\u3002\u4f7f\u7528RNN\u8fdb\u884c\u5efa\u6a21\uff0c\u5176\u8bb0\u5fc6\u673a\u5236\u53ef\u4ee5\u6355\u6349\u5230\u590d\u6742\u7cfb\u7edf\u7684\u957f\u7a0b\u5173\u8054\u3002<\/span><\/section>\n
       <\/span><\/section>\n
      \u57fa\u4e8e RNN \u7684\u65b9\u6cd5\u6709\u975e\u5e38\u591a\uff0c\u8fd9\u91cc\u6211\u4eec\u5c55\u793a\u4e00\u4e2a\u8ddf\u590d\u6742\u7cfb\u7edf\u76f8\u5173\u7684\u5de5\u4f5c\u3002<\/span><\/section>\n
       <\/span><\/section>\n
      \n
      \u5e93\u8ba1\u7b97:<\/span>Model-Free Prediction of Large Spatiotemporally Chaotic Systems from Data: A Reservoir Computing ApproachJaideep Pathak,Brian Hunt,Michelle Girvan.et al.physical review letters\uff082018\uff09<\/span><\/section>\n<\/blockquote>\n
       <\/span><\/section>\n
      \u5728\u8fd9\u4e2a\u5de5\u4f5c\u4e2d\uff0c\u7528 RNN \u7684\u4e00\u79cd\u53d8\u79cd\u6a21\u578b\u201c\u5e93\u8ba1\u7b97\u201d<\/span>\uff08reservoir computing<\/span>\uff09<\/span>\uff0c\u5177\u6709\u5f3a\u5927\u7684\u9884\u6d4b\u80fd\u529b\uff0c\u751a\u81f3\u53ef\u4ee5\u9884\u6d4b\u6df7\u6c8c\u6a21\u578b\u3002\u800c\u5e93\u8ba1\u7b97\u548c\u4e00\u822c\u7684RNN\u6700\u5927\u7684\u533a\u522b\u5728\u4e8e\u5b83\u5177\u6709\u4e00\u4e2a\u4e0d\u53ef\u5b66\u4e60\u7684\u6838\u5fc3\u90e8\u4ef6\uff1a\u6c34\u5e93<\/span>\uff08\u56fe\u4e2d\u7684R\uff09<\/span>\u3002\u6c34\u5e93\u672c\u8d28\u4e0a\u662f\u4e00\u4e2a\u9ad8\u5ea6\u7a00\u758f\u7684\u968f\u673a\u7f51\u7edc\uff0c\u4fe1\u53f7\u7ecf\u8fc7\u7f16\u7801\u540e\u8fdb\u5165\u6c34\u5e93\u8fdb\u884c\u4f20\u9012\uff0c\u800c\u6574\u4e2a\u8fc7\u7a0b\u662f\u4e0d\u53ef\u5b66\u4e60\u7684\u3002\u5728\u8fd9\u4e2a\u8fc7\u7a0b\u4e2d\uff0c\u6c34\u5e93\u8d77\u5230\u8bb0\u5fc6\u7684\u4f5c\u7528\uff0c\u7ecf\u8fc7\u82e5\u5e72\u6b21\u8fed\u4ee3\uff0c\u4fe1\u53f7\u7684\u8f93\u5165\u548c\u8f93\u51fa\u8fc7\u7a0b\u901a\u8fc7\u68af\u5ea6\u7b97\u6cd5\u8fdb\u884c\u5b66\u4e60\u8c03\u6574\u3002\u6700\u540e\uff0c\u8fd9\u4e2a\u6a21\u578b\u5bf9\u4e8e\u6df7\u6c8c\u7684\u9884\u6d4b\uff0c\u5374\u53ef\u4ee5\u5f97\u5230\u4e00\u4e2a\u6bd4LSTM\u7b49\u5b8c\u5168\u68af\u5ea6\u53cd\u5411\u4f20\u64ad\u7684RNN\u66f4\u597d\u7684\u6548\u679c\uff0c\u800c\u5e93\u8ba1\u7b97\u7684\u6838\u5fc3\u90e8\u4ef6\u6c34\u5e93\u7531\u4e8e\u662f\u4e00\u79cd\u56fa\u5b9a\u7ed3\u6784\uff0c\u6211\u4eec\u751a\u81f3\u53ef\u4ee5\u7528\u4e00\u4e9b\u7269\u7406\u4e0a\u7684\u786c\u4ef6\u6765\u5b9e\u73b0\u5b83\u3002<\/span><\/section>\n
       <\/span>  <\/span><\/section>\n
      <\/span><\/section>\n
      \u56fe6\uff1a\u5e93\u8ba1\u7b97\uff08reservoir computing\uff09<\/span><\/section>\n
       <\/span><\/section>\n

      \n
      \u9636\u6bb52\uff1a\u5bf9\u590d\u6742\u7cfb\u7edf\u7684\u52a8\u529b\u5b66\u548c\u63a7\u5236\u81ea\u52a8\u5efa\u6a21\u2014\u2014GNN<\/span><\/strong><\/section>\n<\/h3>\n
                    <\/span><\/section>\n
      <\/section>\n
      \u56fe7\uff1a\u7b2c\u4e09\u4ee3\u795e\u7ecf\u7f51\u7edc\u2014\u2014\u56fe\u795e\u7ecf\u7f51\u7edc<\/span><\/section>\n
       <\/span><\/section>\n
      \u5728\u65f6\u95f4\u5e8f\u5217\u81ea\u52a8\u5b66\u4e60\u65b9\u9762\uff0c\u56fe\u7f51\u7edc\u4e0d\u4ec5\u5f88\u597d\u5730\u5b66\u4e60\u4e86\u52a8\u529b\u5b66\uff0c\u800c\u4e14\u8fd8\u80fd\u591f\u5728\u5b66\u597d\u7684\u52a8\u529b\u5b66\u4e0a\u505a\u63a7\u5236\u3002\u5982\u679c\u5728\u795e\u7ecf\u7f51\u7edc\u4e2d\u5bf9\u56fe\u7f51\u7edc\u8fdb\u884c\u5212\u5206\uff0c\u53ef\u4ee5\u8ba4\u4e3a\u5b83\u662f\u7b2c\u4e09\u4ee3\u7684\u795e\u7ecf\u7f51\u7edc\u3002\u6211\u4eec\u77e5\u9053\uff0c\u7b2c\u4e00\u4ee3\u7684\u795e\u7ecf\u7f51\u7edc\u662f\u5e38\u89c1\u7684\u524d\u9988\u795e\u7ecf\u7f51\u7edc<\/span>\uff08Feed Forward Network\uff09<\/span>\uff0c\u800c\u7b2c\u4e8c\u4ee3\u7684\u795e\u7ecf\u7f51\u7edc\u5219\u662f\u6839\u636e\u95ee\u9898\u8fdb\u884c\u6784\u9020\u7684\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\u3002\u5982\u5728\u56fe\u50cf\u8bc6\u522b\u4e2d\uff0c\u7531\u4e8e\u56fe\u7247\u7ed3\u6784\u4e0a\u7684\u5e73\u79fb\u4e0d\u53d8\u6027\u3001\u65cb\u8f6c\u5bf9\u79f0\u7b49\uff0c\u6211\u4eec\u901a\u8fc7kernel\u5728\u56fe\u7247\u4e0a\u5e73\u79fb\uff0c\u901a\u8fc7\u5377\u79ef\u7684\u65b9\u5f0f\u6765\u63d0\u53d6\u7279\u5f81\uff0c\u7531\u6b64\u8bbe\u8ba1\u4e86CNN\u3002\u800ckernel\u65e0\u8bba\u79fb\u52a8\u5230\u56fe\u7247\u7684\u54ea\u4e00\u4e2a\u4f4d\u7f6e\uff0c\u5176\u5185\u90e8\u7684\u7ed3\u6784\u90fd\u662f\u4e00\u6837\u7684\uff0c\u56e0\u6b64CNN\u53ef\u4ee5\u5b9e\u73b0\u53c2\u6570\u5171\u4eab\u3002\u53c8\u5982\u5728\u81ea\u7136\u8bed\u8a00\u4e2d\uff0c\u6839\u636e\u5176\u4e00\u7ef4\u7ed3\u6784\u7684\u5e8f\u5217\u7279\u6027\uff0c\u901a\u8fc7\u5404\u79cd\u95e8\u7684\u64cd\u4f5c\u4ece\u800c\u4f7f\u5f97\u5e8f\u5217\u524d\u540e\u7684\u4fe1\u606f\u4e92\u76f8\u5f71\u54cd\u4ee5\u66f4\u597d\u6355\u6349\u5e8f\u5217\u7684\u7279\u5f81\uff0c\u7531\u6b64\u8bbe\u8ba1\u4e86RNN\u3002\u800c\u7b2c\u4e09\u4ee3\u795e\u7ecf\u7f51\u7edc\uff0c\u5219\u53ef\u4ee5\u7406\u89e3\u4e3a\u5bf9\u4e0a\u8ff0\u60c5\u51b5\u7684\u4e00\u79cd\u6cdb\u5316\uff0cCNN\u65e0\u975e\u662f\u4e00\u79cd\u56fa\u5b9a\u7684\u82e5\u5e72\u7ef4\u7684\u7f51\u7edc\u7ed3\u6784\uff0c\u800c\u5982\u679c\u5c06\u5176\u6cdb\u5316\u5230\u4e00\u822c\u6027\u7684\u7f51\u7edc\u7ed3\u6784\uff0c\u53bb\u6389\u7a7a\u95f4\u4e0a\u7684\u7ea6\u675f\uff0c\u8fd9\u65f6\u4fbf\u6784\u6210\u4e86\u56fe\u795e\u7ecf\u7f51\u7edc\u3002<\/span><\/section>\n
       <\/span><\/section>\n
      \u56fe\u795e\u7ecf\u7f51\u7edc\u6700\u521d\u7531 Franco Scarselli \u548c Marco Gori \u7b49\u4eba\u63d0\u51fa\uff0c\u5728\u4e4b\u540e\u7684\u5341\u51e0\u5e74\u91cc\u88ab\u4e0d\u65ad\u6269\u5c55\uff0c\u5148\u540e\u53d1\u5c55\u51fa\u4e86\u56fe\u5377\u79ef\u7f51\u7edc<\/span>\uff08Graph Convolution Networks\uff0cGCN\uff09<\/span>\u3001 \u56fe\u6ce8\u610f\u529b\u7f51\u7edc<\/span>\uff08Graph Attention Networks\uff09<\/span>\u3001\u56fe\u81ea\u7f16\u7801\u5668<\/span>\uff08 Graph Autoencoders\uff09<\/span>\u3001\u56fe\u751f\u6210\u7f51\u7edc<\/span>\uff08Graph Generative Networks\uff09<\/span> \u548c\u56fe\u65f6\u7a7a\u7f51\u7edc<\/span>\uff08Graph Spatial-temporal Networks\uff09<\/span>\u7b49\u591a\u4e2a\u5b50\u9886\u57df\u3002<\/span><\/section>\n
       <\/span><\/section>\n
      \u56fe\u7f51\u7edc\u7684\u57fa\u672c\u7406\u5ff5\u662f\u6784\u9020\u4e00\u79cd\u53ef\u5b66\u4e60\u7684\u56fe\u4e0a\u7684\u4f20\u64ad\u8fc7\u7a0b\uff0c\u901a\u8fc7\u7ed9\u673a\u5668\u5c55\u793a\u4e00\u5f20\u56fa\u5b9a\u7684\u56fe\uff0c\u8fd9\u88ab\u79f0\u4e3a\u57fa\u4e8e\u5173\u7cfb\u7684\u504f\u7f6e\u5148\u9a8c<\/span>\uff08inductive bias\uff09<\/span>\uff0c\u5b66\u4e60\u4e00\u79cd\u7f51\u7edc\u8282\u70b9\u5230\u8fde\u8fb9\uff0c\u8fde\u8fb9\u5230\u8282\u70b9\u7684\u6620\u5c04\u8fc7\u7a0b\u3002<\/span><\/section>\n
       <\/span>          <\/span><\/section>\n
      <\/span><\/section>\n
      \u56fe8\uff1a\u56fe\u7f51\u7edc\u52a8\u529b\u5b66\u9884\u6d4b\u539f\u7406<\/span><\/section>\n
       <\/span><\/section>\n
      \u7531\u4e8e\u6a21\u578b\u5c06\u56fe\u7ed3\u6784\u5145\u5206\u5229\u7528\uff0c\u80fd\u591f\u628a\u63e1\u975e\u5e38\u5f02\u8d28\u5316\u7684\u7f51\u7edc\uff0c\u56e0\u6b64\u51c6\u786e\u5ea6\u53ef\u4ee5\u5927\u5927\u9ad8\u3002\u800c\u56fe\u4e0a\u6784\u5efa\u4f20\u64ad\u6269\u6563\u7684\u8fc7\u7a0b\u5373\u4e3a\u4e0d\u540c\u8282\u70b9\u4e4b\u95f4\u4f20\u64ad\u8fc7\u7a0b\u4e2d\u7684\u591a\u6b21\u8fed\u4ee3\uff0c\u901a\u8fc7\u53cd\u5411\u4f20\u64ad\u7b97\u6cd5\u591a\u6b21\u5bf9\u8fde\u8fb9\u4f20\u64ad\u6269\u6563\u8fc7\u7a0b\u4e2d\u7684\u53c2\u6570\u8fdb\u884c\u8c03\u6574\u3002<\/span><\/section>\n
       <\/span>
      <\/section>\n
      \u4e0b\u9762\u8fd9\u7bc7\u6587\u7ae0\u662f\u5f69\u4e91\u5929\u6c14\u7b97\u6cd5\u5de5\u7a0b\u5e08\u548c\u6211\u4eec\u7684\u4e00\u9879\u5de5\u4f5c\uff0c\u6211\u4eec\u5c06\u56fe\u7f51\u7edc\u5e94\u7528\u4e8e\u96fe\u973e\u9884\u6d4b\uff0c\u5e76\u7ed3\u5408\u6c14\u8c61\u9886\u57df\u77e5\u8bc6\uff0c\u6784\u5efa\u57ce\u5e02\u95f4\u7684\u76f8\u4e92\u4f5c\u7528\u7f51\u7edc\u6a21\u578b\uff0c\u5bf9\u9884\u6d4b\u96fe\u973e\u6709\u975e\u5e38\u597d\u7684\u6548\u679c\u3002<\/span><\/section>\n
       <\/span><\/section>\n
      \n
      PM2.5-GNN: A Domain Knowledge Enhanced Graph Neural Network For PM2.5 ForecastingShuo Wang,Yanran Li,Jiang Zhang.et al.arXiv\uff082020\uff09<\/section>\n<\/blockquote>\n
       <\/span><\/section>\n

      \n
      \u56fe\u5377\u79ef\u7f51\u7edc<\/span><\/strong><\/span><\/section>\n<\/h4>\n
       <\/span><\/section>\n
      \u56fe\u5377\u79ef\u7f51\u7edc<\/span>\uff08GCN\uff09<\/span>\u6700\u65e9\u7531Thomas Kipf\u63d0\u51fa<\/span>\uff08\u4eca\u5e74 1 \u6708\u4efd\uff0c\u4ed6\u4ece\u963f\u59c6\u65af\u7279\u4e39\u5927\u5b66\u673a\u5668\u5b66\u4e60\u4e13\u4e1a\u62ff\u5230\u4e86\u535a\u58eb\u5b66\u4f4d\uff09<\/span>\u3002<\/span><\/section>\n
                   <\/span><\/section>\n
      <\/span><\/section>\n
      \u56fe9\uff1aThomas Kipf\u5728\u793e\u4ea4\u5a92\u4f53\u4e0a\u5c55\u793a\u81ea\u5df1\u7684\u535a\u58eb\u8bba\u6587<\/span><\/section>\n

      <\/span><\/section>\n
      \n
      Thomas Kipf \u63d0\u51fa\u4e86\u4e00\u79cd\u7528\u6df1\u5ea6\u5b66\u4e60\u5904\u7406\u7ed3\u6784\u5316\u6570\u636e\u7684\u65b0\u65b9\u6cd5\u3002\u8be5\u65b9\u6cd5\u4e3b\u8981\u57fa\u4e8e\u4ee5\u56fe\u7684\u5f62\u5f0f\u7ed3\u6784\u5316\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u7684\u8868\u793a\u548c\u8ba1\u7b97\uff0c\u4ece\u800c\u63d0\u9ad8\u6a21\u578b\u5728\u5b66\u4e60\u5177\u5907\u663e\u5f0f\u548c\u9690\u5f0f\u6a21\u5757\u7ed3\u6784\u7684\u6570\u636e\u65f6\u7684\u6cdb\u5316\u80fd\u529b\u3002<\/span><\/section>\n<\/blockquote>\n
       <\/span><\/section>\n
      GCN\u7684\u6982\u5ff5\u9996\u6b21\u63d0\u51fa\u4e8eICLR2017<\/span>\uff08\u6210\u6587\u4e8e2016\u5e74\uff09<\/span>\uff0c\u5b83\u5b9e\u9645\u4e0a\u8ddfCNN\u7684\u4f5c\u7528\u4e00\u6837\uff0c\u5c31\u662f\u4e00\u4e2a\u7279\u5f81\u63d0\u53d6\u5668\uff0c\u53ea\u4e0d\u8fc7\u5b83\u7684\u5bf9\u8c61\u662f\u56fe\u6570\u636e\uff0c\u7528\u4e8e\u6267\u884c\u56fe\u7ed3\u6784\u6570\u636e\u4e2d\u8282\u70b9\u7684\u534a\u76d1\u7763\u5206\u7c7b\u4efb\u52a1\u3002GCN\u7cbe\u5999\u5730\u8bbe\u8ba1\u4e86\u4e00\u79cd\u4ece\u56fe\u6570\u636e\u4e2d\u63d0\u53d6\u7279\u5f81\u7684\u65b9\u6cd5\uff0c\u4ece\u800c\u8ba9\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528\u8fd9\u4e9b\u7279\u5f81\u53bb\u5bf9\u56fe\u6570\u636e\u8fdb\u884c\u8282\u70b9\u5206\u7c7b<\/span>\uff08node classification\uff09<\/span>\u3001\u56fe\u5206\u7c7b<\/span>\uff08graph classification\uff09<\/span>\u3001\u8fb9\u9884\u6d4b<\/span>\uff08link prediction\uff09<\/span>\uff0c\u8fd8\u53ef\u4ee5\u987a\u4fbf\u5f97\u5230\u56fe\u7684\u5d4c\u5165\u8868\u793a<\/span>\uff08graph embedding\uff09<\/span>\u3002GCN \u53d1\u8868\u65f6\uff0c\u5b83\u5728\u591a\u4e2a\u65e0\u5411\u56fe\u6570\u636e\u96c6\u7684\u8282\u70b9\u7ea7\u5206\u7c7b\u4efb\u52a1\u4e2d\u5b9e\u73b0\u4e86 SOTA \u6027\u80fd\u3002<\/span><\/section>\n
                   <\/span><\/section>\n
      <\/span><\/section>\n
      \u56fe10\uff1a\u56fe\u5377\u79ef\u7f51\u7edc\u539f\u7406<\/span><\/section>\n
       <\/span><\/section>\n
      \u5176\u4e2dH\u662f\u8282\u70b9\u5411\u91cf\uff0c\u4e5f\u5373\u8282\u70b9\u7279\u5f81\uff0c\u901a\u8fc7\u4e00\u4e2a\u53ef\u5b66\u4e60\u4e00\u822c\u5f62\u5f0fW\u8fdb\u884c\u7ebf\u6027\u53d8\u6362\uff0c\u548c\u4e00\u4e2a\u7684\u6269\u6563\u7b97\u5b50A\uff0c\u518d\u52a0\u4e0a\u4e00\u4e2a\u975e\u7ebf\u6027sigma\u51fd\u6570\uff0c\u8282\u70b9\u7279\u5f81\u5c31\u88ab\u8fed\u4ee3\u4e86\u4e00\u8f6e\uff0c \u800c\u7ecf\u8fc7\u82e5\u5e72\u8f6e\u8fed\u4ee3\uff0c \u5c31\u53ef\u4ee5\u5f97\u5230\u6574\u5408\u4e86\u6574\u4e2a\u7f51\u7edc\u4fe1\u606f\u7684\u8282\u70b9\u8868\u793a\u3002\u800c\u8fd9\u6574\u4e2a\u8fc7\u7a0b\u975e\u5e38\u7c7b\u4f3c\u4e8e\u4e00\u4e2a\u5f52\u4e00\u5316\u7684\u62c9\u666e\u62c9\u65af\u7b97\u5b50\uff0c\u7531\u6b64\u5b8c\u6210\u6574\u4e2a\u56fe\u4e0a\u7684\u6269\u6563\u8fc7\u7a0b\u3002<\/span><\/section>\n

      <\/section>\n

      \n
      \u56fe\u6ce8\u610f\u529b\u7f51\u7edc<\/span><\/strong><\/section>\n<\/h4>\n
       <\/span><\/section>\n
      \u76ee\u524d\u8fd8\u6709\u4e00\u79cd\u5728\u8868\u73b0\u4e0a\u6bd4GCN\u66f4\u52a0\u4f18\u79c0\u7684\u7b97\u6cd5\uff1a\u56fe\u6ce8\u610f\u529b\u7f51\u7edc<\/span>\uff08Graph Attention Network\uff0cGAT\uff09<\/span>\u3002Petar Veli\u010dkovi\u0107<\/span>\uff08\u4e8e 2019 \u5e74\u4ece\u5251\u6865\u5927\u5b66\u83b7\u5f97\u8ba1\u7b97\u673a\u79d1\u5b66\u535a\u58eb\u5b66\u4f4d\uff09<\/span>\u548c Guillem Cucurull\u3001Yoshua Bengio \u7b49\u4eba\u4e00\u8d77\u5b8c\u6210\u4e86\u56fe\u6ce8\u610f\u529b\u7f51\u7edc\u7684\u5f00\u5c71\u4e4b\u4f5c\u2014\u2014\u300aGraph Attention Networks\u300b\uff0c\u8fd9\u7bc7\u8bba\u6587\u88ab ICLR 2018 \u63a5\u6536\u3002Petar Veli\u010dkovi\u0107\u73b0\u5728\u662f DeepMind \u7684\u7814\u7a76\u79d1\u5b66\u5bb6\u3002<\/span><\/section>\n
             <\/span>      <\/span><\/section>\n
      <\/span><\/section>\n
      \u56fe11\uff1aPetar Veli\u010dkovi\u0107\u5728\u793e\u4ea4\u5a92\u4f53\u4e0a\u5c55\u793a\u81ea\u5df1\u7684\u535a\u58eb\u8bba\u6587<\/span><\/section>\n

      <\/span><\/section>\n
      \n
      Petar Veli\u010dkovi\u0107\u7684\u535a\u58eb\u8bba\u6587\u300aThe resurgence of  structure in deep neural networks\u300b\uff0c\u4f5c\u8005\u901a\u8fc7\u5f00\u53d1\u4e09\u79cd structure\u2010infused \u795e\u7ecf\u7f51\u7edc\u67b6\u6784\uff08\u5728\u7a00\u758f\u591a\u6a21\u6001\u548c\u56fe\u7ed3\u6784\u6570\u636e\u4e0a\u8fd0\u884c\uff09\u548c\u4e00\u79cd structure\u2010informed \u56fe\u795e\u7ecf\u7f51\u7edc\u5b66\u4e60\u7b97\u6cd5\u6765\u76f4\u63a5\u9a8c\u8bc1\u8be5\u5047\u8bbe\uff0c\u5e76\u5c55\u793a\u4e86\u4f20\u7edf\u57fa\u7ebf\u6a21\u578b\u548c\u7b97\u6cd5\u7684\u5353\u8d8a\u6027\u80fd\u3002<\/span><\/section>\n<\/blockquote>\n
       <\/span><\/section>\n

      \u9488\u5bf9GCN\u5728\u770b\u5f85\u76f8\u90bb\u8282\u70b9\u7684\u5f71\u54cd\u7684\u65f6\u5019\u8fdb\u884c\u7b80\u5355\u5730\u76f4\u63a5\u5bf9\u8282\u70b9\u7279\u5f81\u8fdb\u884c\u5e73\u6743\u5730\u52a0\u548c\uff0c\u7136\u540e\u6c47\u805a\u5230\u8282\u70b9\u4e0a\uff0c\u5f97\u5230\u65b0\u7684\u7279\u5f81\u8fd9\u4e00\u505a\u6cd5<\/span>\uff0c\u56fe\u6ce8\u610f\u529b\u7f51\u7edc\u63d0\u51fa\u5728\u4f20\u64ad\u8fc7\u7a0b\u5f15\u5165\u81ea\u6ce8\u610f\u529b<\/span>\uff08self-attention\uff09<\/span>\u673a\u5236\uff0c\u6bcf\u4e2a\u8282\u70b9\u7684\u9690\u85cf\u72b6\u6001\u901a\u8fc7\u6ce8\u610f\u5176\u90bb\u5c45\u8282\u70b9\u6765\u8ba1\u7b97\u3002GAT\u7f51\u7edc\u7531\u5806\u53e0\u7b80\u5355\u7684\u56fe\u6ce8\u610f\u529b\u5c42<\/span>\uff08graph attention layer\uff09<\/span>\u6765\u5b9e\u73b0\uff0c\u5bf9\u8282\u70b9\u5bf9  \uff0c\u6ce8\u610f\u529b\u7cfb\u6570\u8ba1\u7b97\u65b9\u5f0f\u5982\u4e0b\u3002<\/span><\/p>\n

             <\/span>      <\/span><\/section>\n
      <\/span><\/section>\n
      <\/span><\/section>\n
      \u56fe12\uff1a\u56fe\u6ce8\u610f\u529b\u7f51\u7edc\u539f\u7406<\/span><\/section>\n
             <\/span><\/span><\/section>\n

      <\/section>\n