{"id":22756,"date":"2020-12-18T21:56:57","date_gmt":"2020-12-18T13:56:57","guid":{"rendered":"https:\/\/swarma.org\/?p=22756"},"modified":"2020-12-18T21:56:57","modified_gmt":"2020-12-18T13:56:57","slug":"%e3%80%8a%e5%9b%a0%e6%9e%9c%e7%a7%91%e5%ad%a6%e5%91%a8%e5%88%8a%e3%80%8b%e7%ac%ac6%e6%9c%9f%ef%bc%9a%e9%a2%86%e5%9f%9f%e8%87%aa%e9%80%82%e5%ba%94","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=22756","title":{"rendered":"\u300a\u56e0\u679c\u79d1\u5b66\u5468\u520a\u300b\u7b2c6\u671f\uff1a\u9886\u57df\u81ea\u9002\u5e94"},"content":{"rendered":"
\n
\n
\n
\n

\u4e3a\u4e86\u5e2e\u52a9\u5927\u5bb6\u66f4\u597d\u5730\u4e86\u89e3\u56e0\u679c\u79d1\u5b66\u7684\u6700\u65b0\u79d1\u7814\u8fdb\u5c55\u548c\u8d44\u8baf\uff0c\u6211\u4eec\u56e0\u679c\u79d1\u5b66\u793e\u533a\u56e2\u961f\u672c\u5468\u6574\u7406\u4e86\u7b2c6\u671f\u300a\u56e0\u679c\u79d1\u5b66\u5468\u520a\u300b\uff0c\u63a8\u9001\u8fd1\u671f\u56e0\u679c\u79d1\u5b66\u503c\u5f97\u5173\u6ce8\u7684\u8bba\u6587\u548c\u8d44\u8baf\u4fe1\u606f\u3002\u672c\u671f\u7684\u4e3b\u9898\u662f\u201d\u9886\u57df\u81ea\u9002\u5e94\u201c\u3002<\/span><\/p>\n

\u672c\u671f\u4f5c\u8005\uff1a\u5218\u73c8\u9e9f \u8bb8\u96c4\u9510 \u8d75\u6c5f\u6770 \u9648\u6657\u66e6 \u8881\u857e \u65b9\u6587\u6bc5 \u6768\u4e8c\u8336 \u9f9a\u9e64\u626c \u5bab\u660e\u660e<\/em><\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n

\n


<\/p>\n

\u5f53\u524d\u673a\u5668\u5b66\u4e60\u7814\u7a76\u7684\u4e00\u4e2a\u91cd\u70b9\uff0c\u662f\u7b97\u6cd5\u7684\u9c81\u68d2\u6027\u548c\u6cdb\u5316\u80fd\u529b\u3002\u5df2\u6709\u6a21\u578b\u5747\u5047\u8bbe\u8bad\u7ec3\u6570\u636e\u548c\u6d4b\u8bd5\u6570\u636e\u72ec\u7acb\u540c\u5206\u5e03\uff08IID\uff09\uff0c\u5b9e\u9645\u5e94\u7528\u5f88\u96be\u6ee1\u8db3\u3002\u66f4\u5e38\u89c1\u7684\u60c5\u51b5\u662f\uff0c\u8bad\u7ec3\u6570\u636e\u548c\u7b97\u6cd5\u90e8\u7f72\u573a\u666f\u7684\u6570\u636e\u4e4b\u95f4\u5b58\u5728\u4e00\u5b9a\u56e0\u679c\u5173\u8054\uff0c\u5bf9\u6b64\uff0cBernhard Scholkopf\u66fe\u4e3e\u4f8b\uff1a\u6d77\u62d4\u662f\u5f71\u54cd\u6c14\u6e29\u7684\u91cd\u8981\u56e0\u7d20\uff0c\u4e0d\u540c\u5730\u533a\u7684\u6d77\u62d4\u4e0d\u540c<\/span>\uff08\u5373\u76ee\u6807\u5148\u9a8c\u77e5\u8bc6\u53d1\u751f\u6539\u53d8\uff09<\/span>\uff0c\u4f46\u6d77\u62d4\u5bf9\u6c14\u6e29\u7684\u5f71\u54cd\u673a\u5236\u51e0\u4e4e\u76f8\u540c<\/span>\uff08\u5373\u6761\u4ef6\u6982\u7387\u5206\u5e03\u76f8\u540c\uff0c\u6216\u56e0\u679c\u673a\u5236\u4e0d\u53d8\uff09<\/span>\u3002\u5929\u4e0b\u6ca1\u6709\u514d\u8d39\u7684\u5348\u9910<\/span>\uff08No Free Launch Theorem\uff09<\/span>\uff0c\u5bf9\u4e8e\u5982\u4f55\u5bfb\u627e\u66f4\u5408\u7406\u7684\u5047\u8bbe\u6765\u4ee3\u66ffIID\u7684\u5f3a\u5047\u8bbe\uff0c\u56e0\u679c\u79d1\u5b66\u4e3a\u6211\u4eec\u63d0\u4f9b\u4e86\u503c\u5f97\u501f\u9274\u7684\u601d\u8def\u3002\u5728\u7b2c\u4e94\u671f\u7684\u5468\u520a\u5f53\u4e2d\uff0c\u6211\u4eec\u4ecb\u7ecd\u4e86\u4e24\u79cd\u89e3\u51b3OOD\u7684\u65b9\u5f0f<\/span>\uff08\u5229\u7528\u56e0\u679c\u56fe\u4fe1\u606f\uff0c\u5229\u7528\u8868\u793a\u5b66\u4e60\uff0cstable learning\u662f\u8fd9\u4e24\u79cd\u601d\u8def\u7684\u878d\u5408\uff09<\/span>\uff0c\u4ee5\u53ca Y. Bengio \u5728\u5c1d\u8bd5\u7684\u5143\u5b66\u4e60\u601d\u8def\u3002\u672c\u671f\u5468\u520a\u4e3a\u5927\u5bb6\u5e26\u6765\uff0c\u7531\u58a8\u5c14\u672c\u5927\u5b66\u5bab\u660e\u660e\u8001\u5e08\u63a8\u8350\u7684\u516d\u7bc7\u5173\u4e8e\u56e0\u679c\u4e0e\u9886\u57df\u81ea\u9002\u5e94\u76f8\u7ed3\u5408\u7684\u6587\u7ae0\u3002\u4e25\u683c\u8303\u7574\u7684 OOD \u95ee\u9898\u8981\u6c42\u6211\u4eec\u4e0d\u5bf9\u6d4b\u8bd5\u6570\u636e\u7684\u5206\u5e03\u505a\u4efb\u4f55\u5047\u8bbe\uff0c\u76f8\u5bf9\u4e8e\u4ece\u7b2c\u4e94\u671f\u5468\u520a\u4ecb\u7ecd\u7684\u6709\u5173 \u4ece IID \u5230 OOD \u7684\u5185\u5bb9\uff0c\u9886\u57df\u81ea\u9002\u5e94\u5173\u6ce8\u7b97\u6cd5\u5728\u6d4b\u8bd5\u96c6\u4e0a\u7684\u8868\u73b0\uff0c\u7ecf\u5e38\u5bf9\u6d4b\u8bd5\u5206\u5e03\u5982\u4f55\u53d8\u5316\u505a\u51fa\u4e86\u4e00\u4e9b\u5047\u5b9a\uff0c\u4f8b\u5982\u76ee\u6807\u504f\u79fb\uff0c\u6761\u4ef6\u504f\u79fb\u548c\u5e7f\u4e49\u76ee\u6807\u504f\u79fb\u3002\u672c\u671f\u5468\u520a\u63a8\u8350\u7684\u516d\u7bc7\u6587\u7ae0\u63a2\u8ba8\u4e86\u8fd9\u4e09\u79cd\u95ee\u9898\uff0c\u5e0c\u671b\u80fd\u6fc0\u53d1\u76f8\u5173\u7814\u7a76\u4eba\u5458\u8fdb\u4e00\u6b65\u7684\u601d\u8003\u3002<\/span><\/p>\n


<\/section>\n

1\uff09Zhang, Kun, et al. “Domain adaptation under target and conditional shift.” International Conference on Machine Learning. 2013.<\/span><\/p>\n

2\uff09Gong, Mingming, et al. “Domain adaptation with conditional transferable components.” International conference on machine learning. 2016.<\/span><\/p>\n

3\uff09Guo, Jiaxian, et al. “Ltf: A label transformation framework for correcting target shift.” ICML, 2020.<\/span><\/p>\n

4\uff09Teshima, Takeshi, Issei Sato, and Masashi Sugiyama. “Few-shot Domain Adaptation by Causal Mechanism Transfer.” arXiv preprint arXiv:2002.03497 (2020).
<\/span><\/p>\n

5\uff09Magliacane, Sara, et al. “Domain adaptation by using causal inference to predict invariant conditional distributions.” Advances in Neural Information Processing Systems. 2018.
<\/span><\/p>\n

6\uff09Zhang, Kun, et al. “Domain adaptation as a problem of inference on graphical models.” Advances in Neural Information Processing Systems 33 (2020).<\/span>
<\/p>\n


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

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

<\/section>\n

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

1. \u8bba\u6587\u7ffb\u8bd1\u548c\u89e3\u8bfb<\/span><\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n
 <\/span>
<\/section>\n
\u6211\u4eec\u63a8\u8350\u76846\u7bc7\u8bba\u6587\u6839\u636e\u56e0\u679c\u6a21\u578b\u662f\u5426\u5df2\u77e5\u53ef\u4ee5\u5206\u4e3a\u4e24\u5927\u7c7b\uff0c\u524d4\u7bc7\u662f\u5728\u56e0\u679c\u6a21\u578b\u5df2\u77e5\u7684\u6761\u4ef6\u4e0b\uff0c\u5176\u4e2d\u524d3\u7bc7\u6587\u7ae0\u63a2\u8ba8\u4e86\u5728\u56e0\u679c\u56fe\u5df2\u77e5\u7684\u60c5\u51b5\u4e0b\uff0c\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u7684\u57fa\u672c\u6846\u67b6\uff0c\u5e7f\u4e49\u76ee\u6807\u504f\u79fb\u548c\u8868\u793a\u5b66\u4e60\u76f8\u7ed3\u5408\u7684\u76ee\u6807\u504f\u79fb\u95ee\u9898\uff0c\u7b2c4\u7bc7\u8bba\u6587\u5219\u5229\u7528\u5230\u4e86J.Pearl\u7684\u7ed3\u6784\u56e0\u679c\u65b9\u7a0bSCM\u6765\u5e2e\u52a9\u9886\u57df\u81ea\u9002\u5e94\uff1b\u7b2c5\u30016\u7bc7\u8bba\u6587\u5219\u5c5e\u4e8e\u56e0\u679c\u6a21\u578b\u672a\u77e5\u7684\u60c5\u51b5\u3002<\/span><\/section>\n

<\/span><\/section>\n
1.1 \u56e0\u679c\u6a21\u578b\u5df2\u77e5<\/span><\/strong><\/span><\/section>\n

<\/span><\/section>\n
Zhang, Kun, et al. “Domain adaptation under target and conditional shift.” International Conference on Machine Learning. 2013.<\/span><\/section>\n

<\/span><\/section>\n
\u6458\u8981<\/strong>\uff1a\u7528X\u8868\u793a\u7279\u5f81\uff0cY\u8868\u793a\u76ee\u6807\uff0c\u672c\u6587\u7814\u7a76\u4ee5\u4e0b\u4ee5\u4e0b\u4e09\u79cd\u60c5\u51b5\u7684\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\uff1a\uff081\uff09\u76ee\u6807\u504f\u79fb\uff0c\u8fb9\u7f18\u6982\u7387<\/span>\u6539\u53d8\uff0c\u6761\u4ef6\u6982\u7387<\/span>\u4e0d\u53d8\uff1b\uff082\uff09\u6761\u4ef6\u504f\u79fb\uff0c\u4e0d\u53d8\uff0c\u6761\u4ef6\u6982\u7387\u5728\u9650\u5b9a\u6761\u4ef6\u4e0b\u6539\u53d8\uff1b\uff083\uff09\u5e7f\u4e49\u76ee\u6807\u504f\u79fb\uff0c\u6539\u53d8\uff0c\u6761\u4ef6\u6982\u7387\u5728\u9650\u5b9a\u6761\u4ef6\u4e0b\u6539\u53d8\u3002\u5229\u7528\u80cc\u666f\u56e0\u679c\u77e5\u8bc6\uff0c\u6211\u4eec\u53ef\u4ee5\u5bf9\u624b\u4e2d\u7684\u95ee\u9898\u8fdb\u884c\u6b63\u786e\u7684\u5206\u7c7b\u3002\u6211\u4eec\u5229\u7528\u91cd\u8981\u6027\u91cd\u4f30\u548c\u6837\u672c\u53d8\u6362\u6765\u627e\u5230\u5728\u6d4b\u8bd5\u96c6\u4e0a\u6027\u80fd\u826f\u597d\u7684\u5b66\u4e60\u7b97\u6cd5\uff0c\u601d\u8def\u662f\u5bf9\u8bad\u7ec3\u96c6\u6570\u636e\u8fdb\u884c\u91cd\u65b0\u52a0\u6743\u8bc4\u4f30\u6216\u53d8\u6362\u6765\u91cd\u73b0\u6d4b\u8bd5\u96c6\u4e0a\u7684\u534f\u53d8\u91cf\u5206\u5e03\u3002\u56e0\u4e3a\u8fb9\u7f18\u6982\u7387\u5206\u5e03\u548c\u6761\u4ef6\u6982\u7387\u5206\u5e03\u90fd\u6709\u6838\u65b9\u6cd5\u8868\u793a\uff0c\u672c\u6587\u63d0\u51fa\u7684\u65b9\u6cd5\u907f\u514d\u4e86\u76f4\u63a5\u91cd\u65b0\u4f30\u8ba1\u534f\u53d8\u91cf\u5206\u5e03\uff0c\u4e5f\u9002\u7528\u4e8e\u9ad8\u7ef4\u6570\u636e\u5206\u5e03\u3002\u5728\u4eff\u771f\u6570\u636e\u548c\u771f\u5b9e\u6570\u636e\u4e0a\u7684\u6570\u503c\u5b9e\u9a8c\u8bc1\u660e\u4e86\u672c\u6587\u6240\u63d0\u65b9\u6cd5\u7684\u6709\u6548\u6027\u3002<\/span><\/section>\n

<\/span><\/section>\n
\u8bd1\u8005<\/strong>\uff1a\u5218\u73c8\u9e9f<\/span><\/section>\n
\u5218\u73c8\u9e9f\u7b80\u8bc4\uff1a<\/span><\/strong>Zhang Kun\uff08CMU\uff0cMPI\uff09\u8001\u5e08\u8fd9\u7bc7\u6587\u7ae0\uff0c\u5c06\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u68b3\u7406\u6210\u56db\u4e2a\u7c7b\u522b\uff1a\u534f\u53d8\u91cf\u504f\u79fb\uff0c\u76ee\u6807\u504f\u79fb\uff0c\u6761\u4ef6\u504f\u79fb\u548c\u5e7f\u4e49\u76ee\u6807\u504f\u79fb\u3002\u5e76\u9488\u5bf9\u4e0a\u8ff0\u4e09\u79cd\u672a\u5f97\u5230\u89e3\u51b3\u7684\u95ee\u9898\u7ed9\u51fa\u4e86\u57fa\u672c\u89e3\u51b3\u6846\u67b6\u3002<\/span><\/section>\n

<\/span><\/section>\n
Gong, Mingming, et al. “Domain adaptation with conditional transferable components.” International conference on machine learning. 2016.<\/span><\/section>\n

<\/span><\/section>\n
\u6458\u8981\uff1a<\/strong>\u9886\u57df\u81ea\u9002\u5e94\u5728\u76d1\u7763\u5b66\u4e60\u4e2d\u7528\u4e8e\u7814\u7a76\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u5b58\u5728\u6570\u636e\u5206\u5e03\u4e0d\u4e00\u81f4\u7684\u60c5\u51b5\u3002\u5728\u4e4b\u524d\u7684\u5de5\u4f5c\u4e2d\uff0c\u5927\u90e8\u5206\u7684\u9886\u57df\u81ea\u9002\u5e94\u5de5\u4f5c\u90fd\u96c6\u4e2d\u5728\u534f\u53d8\u91cf\u504f\u79fb\u7684\u7814\u7a76\uff0c \u7528X\u548cY\u5206\u522b\u8868\u793a\u7279\u5f81\u548c\u76ee\u6807\uff0c\u5f53\u51fa\u73b0\u8de8\u9886\u57df\u65f6\uff0c\u4ec5\u8003\u8651\u7279\u5f81\u5206\u5e03P(X)\u7684\u6539\u53d8\uff0c\u5e76\u6ca1\u6709\u8003\u8651\u6761\u4ef6\u6982\u7387P(Y|X)\u7684\u53d8\u5316\u3002\u4e3a\u4e86\u964d\u4f4e\u9886\u57df\u504f\u5dee\uff0c\u6700\u8fd1\u7684\u7814\u7a76\u901a\u8fc7\u6700\u5c0f\u5316\u4e0d\u540c\u9886\u57df\u5206\u5e03\u5dee\u5f02\u7684\u65b9\u6cd5\uff0c\u5bfb\u627e\u4e0d\u53d8\u56e0\u5b50 <\/span>(invariant components)<\/span> T(x)\uff0cT(x)\u7684\u6982\u7387\u5206\u5e03P(T(x))\u5728\u4e0d\u540c\u9886\u57df\u4e2d\u662f\u8fd1\u4f3c\u76f8\u7b49\u7684\u3002\u7136\u800c\u5728\u4e0d\u540c\u7684\u9886\u57df\u4e2d\uff0c\u5f53P(Y|X)\u53d1\u751f\u53d8\u5316\u65f6\uff0c\u5e76\u4e0d\u80fd\u786e\u5b9a\u5b83\u4eec\u7684P(Y|T(X))\u4e5f\u662f\u8fd1\u4f3c\u76f8\u7b49\u7684\u3002\u6b64\u5916\uff0c\u53ef\u8fc1\u79fb\u7684\u56e0\u5b50 <\/span>(transferable components) <\/span>\u5e76\u4e00\u5b9a\u662f\u4e0d\u53d8\u7684\u3002\u53ea\u8981\u5176\u4e2d\u6709\u4e00\u4e9b\u56e0\u5b50\u7684\u53d8\u5316\u662f\u53ef\u8bc6\u522b\u7684\uff0c\u5c31\u53ef\u4ee5\u5229\u7528\u8fd9\u4e9b\u56e0\u5b50\u5728\u76ee\u6807\u9886\u57df\u8fdb\u884c\u9884\u6d4b\u3002\u672c\u6587\u4e3b\u8981\u5173\u6ce8\u5728Y causes X\u7684\u56e0\u679c\u7cfb\u7edf\u4e2d\uff0c\u7814\u7a76P(X|Y)\u548cP(Y)\u90fd\u53d1\u751f\u53d8\u5316\u7684\u60c5\u51b5\u3002\u5728\u5408\u9002\u7684\u5047\u8bbe\u524d\u63d0\u4e0b\uff0c\u6211\u4eec\u7684\u76ee\u6807\u662f\u63d0\u53d6\u90a3\u4e9b\u901a\u8fc7\u5c40\u90e8\u7f29\u653e<\/span> (location-scale) <\/span>\u53d8\u6362\u4e4b\u540e\uff0c\u5728\u4e0d\u540c\u9886\u57df\u5185\u7684\u6761\u4ef6\u6982\u7387P(T(X)|Y)\u4fdd\u6301\u4e0d\u53d8\u7684\u6761\u4ef6\u53ef\u8fc1\u79fb\u56e0\u5b50 <\/span>(conditional transferable components)<\/span>\uff0c\u5e76\u540c\u65f6\u8bc6\u522bP(Y)\u5728\u4e0d\u540c\u9886\u57df\u95f4\u662f\u5982\u4f55\u53d8\u5316\u7684\u3002\u672c\u6587\u5728\u4eba\u5de5\u5408\u6210\u6570\u636e\u4e0e\u771f\u5b9e\u4e16\u754c\u6570\u636e\u4e0a\u8fdb\u884c\u4e86\u7406\u8bba\u5206\u6790\u548c\u5b9e\u8bc1\u8bc4\u4f30\uff0c\u6765\u8868\u660e\u6211\u4eec\u65b9\u6cd5\u7684\u6709\u6548\u6027\u3002<\/span><\/section>\n

<\/span><\/section>\n
\u8bd1\u8005<\/strong>\uff1a\u8d75\u6c5f\u6770\uff0c\u8bb8\u96c4\u9510<\/span><\/section>\n
\u5218\u73c8\u9e9f\u7b80\u8bc4\uff1a<\/span><\/strong>\u672c\u6587\u5728\u7b2c\u4e00\u7bc7\u6587\u7ae0\u7684\u57fa\u7840\u4e0a\uff0c\u8003\u8651\u5e7f\u4e49\u76ee\u6807\u504f\u79fb\u6761\u4ef6\u4e0b\u7684\u9886\u57df\u81ea\u9002\u5e94\u3002\u4e0e\u63a8\u8350\u8bba\u65871\u5173\u6ce8\u6837\u672c\u91cd\u8981\u6027\u8de8\u9886\u57df\u91cd\u4f30\u4e0d\u540c\u7684\u662f\uff0c\u672c\u6587\u5173\u6ce8\u53e6\u4e00\u5927\u7c7b\u65b9\u6cd5\uff1a\u4e0d\u53d8\u56e0\u5b50\u53d1\u73b0\u3002\u5df2\u6709\u7814\u7a76\u5ffd\u7565\u4e86\u4e00\u4e2a\u91cd\u8981\u73b0\u8c61\uff0c\u5373\u4e0d\u53d8\u56e0\u5b50\u53ef\u80fd\u6539\u53d8\uff0c\u672c\u6587\u4f5c\u8005\u9488\u5bf9\u8fd9\u79cd\u60c5\u51b5\u63d0\u51fa\u4e86\u65b0\u7684\u65b9\u6cd5\u6765\u7edf\u4e00\u5df2\u6709\u7684\u4e0d\u53d8\u56e0\u5b50\u53d1\u73b0\u76f8\u5173\u7814\u7a76\u3002<\/span><\/section>\n

<\/span><\/section>\n
Guo, Jiaxian, et al. “Ltf: A label transformation framework for correcting target shift.” ICML, 2020.<\/span><\/section>\n

<\/span><\/section>\n
\u6458\u8981<\/strong>\uff1a\u5206\u5e03\u504f\u79fb\u662f\u5f53\u524d\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u5728\u73b0\u5b9e\u4e16\u754c\u4e2d\u90e8\u7f72\u65f6\u9047\u5230\u7684\u4e3b\u8981\u969c\u788d\u3002\u7528Y\u8868\u793a\u6807\u7b7e\u3001X\u8868\u793a\u7279\u5f81\uff0c\u6211\u4eec\u5173\u6ce8\u7684\u662f\u5206\u5e03\u504f\u79fb\u4e2d\u7684\u4e00\u79cd\u2014\u2014target shift\uff0c\u8fd9\u91cc\u662f\u6307\u76ee\u6807\u53d8\u91cf\u7684\u8fb9\u7f18\u5206\u5e03\u53d1\u751f\u53d8\u5316\u800c\u6761\u4ef6\u5206\u5e03\u4e0d\u53d8\u7684\u60c5\u51b5\u3002\u73b0\u6709\u65b9\u6cd5\u901a\u8fc7\u5bc6\u5ea6\u4f30\u8ba1\u7684\u65b9\u6cd5\u6765\u4f30\u8ba1\u6e90\u57df\u548c\u76ee\u6807\u57df\u7684\u6807\u7b7e\u5206\u5e03\u5bc6\u5ea6\u7387\u3002\u7136\u800c\uff0c\u8fd9\u4e9b\u65b9\u6cd5\u8981\u4e48\u5bf9\u5927\u89c4\u6a21\u6570\u636e\u6709\u5f88\u9ad8\u7684\u8ba1\u7b97\u91cf\uff0c\u8981\u4e48\u9650\u5b9a\u5728\u79bb\u6563\u6807\u7b7e\u504f\u79fb\u7684\u60c5\u51b5\u3002\u672c\u6587\u4e2d\uff0c\u6211\u4eec\u63d0\u51fa\u4e86\u4e00\u79cd\u7aef\u5230\u7aef\u7684\u6807\u7b7e\u8f6c\u6362\u6846\u67b6<\/span> (Label Transformation Framework, LTF) <\/span>\u6765\u4fee\u6b63target shift\uff0cLTF\u628a\u7684\u504f\u79fb\u548c\u6761\u4ef6\u5206\u5e03\u7528\u795e\u7ecf\u7f51\u7edc\u5efa\u6a21\u3002\u7531\u4e8e\u6df1\u5ea6\u7f51\u7edc\u7684\u7075\u6d3b\u6027\uff0c\u6211\u4eec\u63d0\u51fa\u7684\u6846\u67b6\u53ef\u4ee5\u9488\u5bf9\u5927\u6570\u636e\u7528\u7edf\u4e00\u7684\u6846\u67b6\u5904\u7406\u8fde\u7eed\u3001\u79bb\u6563\u751a\u81f3\u591a\u7ef4\u6807\u7b7e\u3002\u6b64\u5916\uff0c\u5bf9\u9ad8\u7ef4\u7279\u5f81X\uff0c\u4f8b\u5982\u56fe\u50cf\uff0c\u6211\u4eec\u53d1\u73b0X\u4e2d\u5197\u4f59\u7684\u4fe1\u606f\u4e25\u91cd\u964d\u4f4e\u4e86\u4f30\u8ba1\u51c6\u786e\u7387\u3002\u4e3a\u4e86\u89e3\u51b3\u8fd9\u4e00\u95ee\u9898\uff0c\u6211\u4eec\u63d0\u51fa\u4e00\u79cd\u57fa\u4e8e\u751f\u6210\u6a21\u578b\u9690\u5f0f\u5730\u5339\u914d\u5206\u5e03\uff0c\u5c06\u7279\u5f81\u5206\u5e03\u5339\u914d\u5230\u4f4e\u7ef4\u7279\u5f81\u7a7a\u95f4\uff0c\u4e14\u4e22\u5f03\u90a3\u4e9b\u548cY\u65e0\u5173\u7684\u4fe1\u606f\u3002\u7406\u8bba\u548c\u7ecf\u9a8c\u7814\u7a76\u90fd\u8868\u660e\u6211\u4eec\u7684\u65b9\u6cd5\u6bd4\u5148\u524d\u7684\u5de5\u4f5c\u6709\u66f4\u5353\u8d8a\u7684\u6027\u80fd\u3002<\/span><\/section>\n

<\/span><\/section>\n
\u8bd1\u8005<\/strong>\uff1a\u9648\u6657\u66e6<\/span><\/section>\n
\u5218\u73c8\u9e9f\u7b80\u8bc4\uff1a<\/span><\/strong>\u672c\u6587\u548c\u5bab\u660e\u660e\u8001\u5e08\uff0cZhang Kun\u8001\u5e08\u7684\u7814\u7a76\u601d\u8def\u4e00\u8109\u76f8\u627f\u3002\u524d\u4e24\u7bc7\u6587\u7ae0\u5206\u522b\u6307\u51fa\u4e86\u9886\u57df\u81ea\u9002\u5e94\u7684\u91cd\u70b9\u4e5f\u662f\u96be\u70b9\uff1a\u4ece\u6e90\u57df\u5230\u76ee\u6807\u57df\u7684\u8fc1\u79fb\u673a\u5236\uff0c\u8fd9\u5728\u6837\u672c\u91cd\u4f30\u4e2d\u662fratio\uff0c\u90fd\u662f\u9488\u5bf9\u6982\u7387\u5bc6\u5ea6\u8fdb\u884c\u76f4\u63a5\u6216\u95f4\u63a5\u7684\u91cd\u4f30\u3002\u4e00\u4e2a\u81ea\u7136\u7684\u60f3\u6cd5\u5c31\u662f\u7528\u795e\u7ecf\u7f51\u7edc\u6765\u4ee3\u66ff\u8fd9\u4e2a\u91cd\u4f30\u8fc7\u7a0b\u3002\u7ed3\u5408System1\u8868\u793a\u5b66\u4e60\u7684\u6210\u679c\u7814\u7a76\u9886\u57df\u81ea\u9002\u5e94\u662f\u672c\u6587\u7684\u4eae\u70b9\u3002<\/span><\/section>\n

<\/span><\/section>\n
Teshima, Takeshi, Issei Sato, and Masashi Sugiyama. “Few-shot Domain Adaptation by Causal Mechanism Transfer.” arXiv preprint arXiv:2002.03497<\/em> (2020).<\/span><\/section>\n

<\/span><\/section>\n
\u6458\u8981<\/strong>\uff1a\u6211\u4eec\u7814\u7a76\u56de\u5f52\u4e2d\u7684\u5c11\u6837\u672c\u6709\u76d1\u7763\u9886\u57df\u81ea\u9002\u5e94<\/span>\uff08DA\uff09<\/span>\u95ee\u9898\uff0c\u5728\u8be5\u95ee\u9898\u4e2d\u6211\u4eec\u53ea\u80fd\u83b7\u5f97\u5c11\u91cf\u7684\u76ee\u6807\u57df\u6709\u6807\u7b7e\u6570\u636e\u548c\u5927\u91cf\u7684\u6e90\u57df\u6709\u6807\u7b7e\u6570\u636e\u3002\u73b0\u5728\u5927\u591a\u6570\u7684DA\u65b9\u6cd5\u90fd\u662f\u5c06\u4ed6\u4eec\u7684\u8fc1\u79fb\u5047\u8bbe\u5efa\u7acb\u5728\u5206\u5e03\u53d8\u5316\u7684\u53c2\u6570\u5f62\u5f0f\u6216\u662f\u660e\u663e\u7684\u5206\u5e03\u76f8\u4f3c\u5904\u4e0a\uff0c\u6bd4\u5982\u8bf4\u76f8\u540c\u7684\u6761\u4ef6\u6216\u662f\u5c0f\u7684\u5206\u5e03\u5dee\u5f02\u3002\u7136\u800c\uff0c\u8fd9\u4e9b\u5047\u8bbe\u53ef\u80fd\u9650\u5236\u4e86\u590d\u6742\u53d8\u5316\u7684\u6216\u662f\u660e\u663e\u975e\u5e38\u4e0d\u540c\u5206\u5e03\u7684\u81ea\u9002\u5e94\u53ef\u80fd\u6027\u3002\u4e3a\u4e86\u514b\u670d\u8fd9\u4e2a\u95ee\u9898\uff0c\u6211\u4eec\u63d0\u51fa\u4e86\u673a\u5236\u8fc1\u79fb\u2014\u2014\u4e00\u79cd\u8de8\u9886\u57df\u65f6\u6570\u636e\u4ea7\u751f\u673a\u5236\u4e0d\u53d8\u7684\u5143\u5206\u5e03\u60c5\u5f62\u3002\u673a\u5236\u8fc1\u79fb\u7684\u5047\u8bbe\u80fd\u591f\u5728\u7ed9DA\u63d0\u4f9b\u575a\u5b9e\u7684\u7edf\u8ba1\u57fa\u7840\u7684\u540c\u65f6\u8003\u8651\u5230\u5bfc\u81f4\u660e\u663e\u4e0d\u540c\u5206\u5e03\u7684\u975e\u53c2\u6570\u53d8\u5316\u60c5\u5f62\u3002\u6211\u4eec\u5c06\u56e0\u679c\u6a21\u578b\u4e2d\u7684\u7ed3\u6784\u65b9\u7a0b\u4f5c\u4e3a\u6848\u4f8b\u5e76\u4e14\u63d0\u51fa\u4e86\u4e00\u79cd\u521b\u65b0\u7684\u4e14\u901a\u8fc7\u7406\u8bba\u548c\u5b9e\u9a8c\u8bc1\u660e\u6709\u7528\u7684DA\u65b9\u6cd5\u3002\u6211\u4eec\u7684\u65b9\u6cd5\u53ef\u4ee5\u88ab\u89c6\u4f5c\u5728DA\u4e2d\u5b8c\u5168\u53d1\u6325\u7ed3\u6784\u56e0\u679c\u6a21\u578b\u7684\u91cd\u8981\u4f5c\u7528\u7684\u7b2c\u4e00\u6b21\u5c1d\u8bd5\u3002<\/span><\/section>\n

<\/span><\/section>\n
\u8bd1\u8005<\/strong>\uff1a\u8881\u857e<\/span><\/section>\n
\u5218\u73c8\u9e9f\u7b80\u8bc4\uff1a<\/span><\/strong>Elias Bareinboim\u5728\u6587\u7ae0\u201cOn Pearl\u2019s Hierarchy and the Foundations of Causal Inference\u201d\u4e2d\u6307\u51fa\uff0c\u56e0\u679c\u63a8\u7406\u672c\u8d28\u4e0a\u662f\u5728SCM\u672a\u77e5\u7684\u60c5\u51b5\u4e0b\uff0c\u901a\u8fc7\u7ecf\u9a8c<\/span>\uff08\u6570\u636e\uff09<\/span>\u548c\u77e5\u8bc6<\/span>\uff08\u56e0\u679c\u5047\u8bbe\u9650\u5236\uff09<\/span>\u6765\u505a\u5206\u7c7b\u56de\u5f52\u7b49\u63a8\u7406\u3002\u672c\u6587\u540c\u6837\u662f\u57fa\u4e8e\u8fd9\u6837\u4e00\u4e2a\u5047\u8bbe<\/span>\uff08\u5728\u6570\u636e\u5206\u5e03\u80cc\u540e\u5b58\u5728\u4e00\u4e2a\u666e\u4e16\u7684\u56e0\u679c\u751f\u6210\u673a\u5236\uff09<\/span>\uff0c\u901a\u8fc7\u5bf9\u8fd9\u4e2a\u751f\u6210\u673a\u5236\u505a\u5047\u8bbe\u53ef\u4ee5\u907f\u514d\u76f4\u63a5\u5bf9\u8fc1\u79fb\u8fc7\u7a0b\u505a\u5047\u8bbe<\/span>\uff08Zhang Kun\u8001\u5e08\u7b2c\u4e00\u7bc7\u6587\u7ae0\u4e2d\u5047\u8bbe\u8fc1\u79fb\u670d\u4ece\u5c40\u90e8\u53d8\u6362\uff09<\/span>\uff0c\u662fSCM\u6846\u67b6\u548c\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u7684\u4e00\u6b21\u7ed3\u5408\u3002<\/span><\/section>\n

<\/span><\/section>\n
1.2 \u56e0\u679c\u6a21\u578b\u672a\u77e5<\/span><\/strong><\/span><\/section>\n

<\/span><\/section>\n
Magliacane, Sara, et al. “Domain adaptation by using causal inference to predict invariant conditional distributions.” Advances in Neural Information Processing Systems<\/em>. 2018.<\/span><\/section>\n

<\/strong><\/span><\/section>\n
\u6458\u8981<\/strong>\uff1a\u9886\u57df\u81ea\u9002\u5e94\u548c\u56e0\u679c\u63a8\u7406\u7684\u4e00\u4e2a\u5171\u540c\u7684\u91cd\u8981\u76ee\u6807\u662f\u5f53\u6e90<\/span>\uff08\u6216\u8bad\u7ec3\uff09<\/span>\u9886\u57df\u548c\u76ee\u6807<\/span>\uff08\u6216\u6d4b\u8bd5\uff09<\/span>\u9886\u57df\u7684\u5206\u5e03\u4e0d\u540c\u65f6\u505a\u51fa\u51c6\u786e\u7684\u9884\u6d4b\u3002\u5728\u8bb8\u591a\u60c5\u51b5\u4e0b\uff0c\u8fd9\u4e9b\u4e0d\u540c\u7684\u5206\u5e03\u53ef\u4ee5\u88ab\u5efa\u6a21\u4e3a\u4e00\u4e2a\u5355\u4e00\u5e95\u5c42\u7cfb\u7edf\u4e0b\u7684\u4e0d\u540c\u73af\u5883\uff0c\u5176\u4e2d\u6bcf\u4e2a\u5206\u5e03\u5bf9\u5e94\u4e8e\u7cfb\u7edf\u7684\u4e0d\u540c\u6270\u52a8\uff0c\u6216\u8005\u4f7f\u7528\u56e0\u679c\u672f\u8bed\u2014\u2014\u5e72\u9884\u3002\u6211\u4eec\u4e13\u6ce8\u4e8e\u8fd9\u6837\u7684\u4e00\u7c7b\u56e0\u679c\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\uff0c\u95ee\u9898\u7ed9\u51fa\u4e86\u4e00\u4e2a\u6216\u591a\u4e2a\u6e90\u57df\u7684\u6570\u636e\uff0c\u4efb\u52a1\u662f\u901a\u8fc7\u5bf9\u4e00\u4e2a\u6216\u591a\u4e2a\u76ee\u6807\u57df\u4e2d\u5176\u4ed6\u53d8\u91cf\u7684\u6d4b\u91cf\u6765\u9884\u6d4b\u67d0\u4e2a\u76ee\u6807\u53d8\u91cf\u7684\u5206\u5e03\u3002\u6211\u4eec\u63d0\u51fa\u4e86\u4e00\u79cd\u89e3\u51b3\u8fd9\u4e9b\u95ee\u9898\u7684\u65b9\u6cd5\uff0c\u8be5\u65b9\u6cd5\u5229\u7528\u56e0\u679c\u63a8\u7406\uff0c\u4e0d\u4f9d\u8d56\u4e8e\u5bf9\u56e0\u679c\u56fe\u3001\u5e72\u9884\u7c7b\u578b\u6216\u5e72\u9884\u76ee\u6807\u7684\u5148\u9a8c\u77e5\u8bc6\u3002\u6211\u4eec\u901a\u8fc7\u5728\u6a21\u62df\u548c\u771f\u5b9e\u6570\u636e\u4e0a\u8bc4\u4f30\u4e00\u4e2a\u53ef\u80fd\u7684\u5b9e\u73b0\u6765\u5c55\u793a\u6211\u4eec\u7684\u65b9\u6cd5\u3002<\/span><\/section>\n

<\/span><\/section>\n
\u8bd1\u8005<\/strong>\uff1a\u65b9\u6587\u6bc5<\/span><\/section>\n
\u9f9a\u9e64\u626c\u7b80\u8bc4\uff1a<\/span><\/strong>\u6587\u7ae0\u7684\u4e00\u4e2a\u6838\u5fc3\u5047\u8bbe\u662f\u5b58\u5728\u4e00\u4e2a\u534f\u53d8\u91cf\u5b50\u96c6 A<\/em>\uff0c\u4f7f\u5f97\u7ed9\u5b9a\u8fd9\u4e9b\u534f\u53d8\u91cf\u7684\u60c5\u51b5\u4e0b\uff0cY <\/em>\u7684\u6761\u4ef6\u5206\u5e03\u5728\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u4e0a\u662f\u4e0d\u53d8\u7684\u3002<\/span><\/section>\n

<\/span><\/section>\n
Zhang, Kun, et al. “Domain adaptation as a problem of inference on graphical models.” Advances in Neural Information Processing Systems<\/em> 33 (2020).<\/span><\/section>\n

<\/span><\/section>\n
\u6458\u8981<\/strong>\uff1a\u672c\u6587\u81f4\u529b\u4e8e\u89e3\u51b3\u6570\u636e\u9a71\u52a8\u7684\u65e0\u76d1\u7763\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u3002\u65e0\u76d1\u7763\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u7684\u96be\u70b9\u5728\u4e8e\u4e8b\u5148\u65e0\u6cd5\u5f97\u77e5\u8054\u5408\u5206\u5e03\u662f\u5982\u4f55\u8de8\u9886\u57df\u53d8\u5316\u7684\uff0c\u4e5f\u5c31\u662f\u8bf4\uff0c\u65e0\u6cd5\u5f97\u77e5\u5728\u4e0d\u540c\u7684\u9886\u57df\u4e2d\u5bf9\u53d8\u91cf\u5206\u5e03\u6709\u4f5c\u7528\u7684\u54ea\u4e9b\u56e0\u7d20\u6216\u6a21\u5757\u662f\u4fdd\u6301\u4e0d\u53d8\u7684\uff0c\u6291\u6216\u662f\u53d8\u5316\u7684\u3002\u4e3a\u4e86\u5f97\u5230\u4e00\u79cd\u89e3\u51b3\u591a\u6e90\u57df\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u7684\u81ea\u52a8\u5316\u65b9\u6cd5\uff0c\u6211\u4eec\u4f7f\u7528\u4e86\u80fd\u4ece\u6570\u636e\u4e2d\u5b66\u4e60\u5230\u7684\u56fe\u6a21\u578b\u4f5c\u4e3a\u7f16\u7801\u8054\u5408\u5206\u5e03\u4e2d\u53d8\u5316\u5c5e\u6027\u7684\u538b\u7f29\u65b9\u6cd5\uff0c\u7136\u540e\u5c06\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u89c6\u4f5c\u5728\u56fe\u6a21\u578b\u4e0a\u7684\u8d1d\u53f6\u65af\u63a8\u65ad\u95ee\u9898\u3002\u8fd9\u6837\u7684\u4e00\u79cd\u56fe\u6a21\u578b\u80fd\u591f\u5c06\u5206\u5e03\u4e2d\u7684\u4e0d\u53d8\u6a21\u5757\u548c\u53d8\u5316\u6a21\u5757\u533a\u5206\u5f00\u6765\uff0c\u5e76\u80fd\u786e\u5b9a\u8de8\u9886\u57df\u7684\u53d8\u5316\u5c5e\u6027\uff0c\u8fd9\u53ef\u4ee5\u4e3a\u63a8\u5bfc\u76ee\u6807\u57df\u4e2d\u76ee\u6807\u53d8\u91cfY\u7684\u540e\u9a8c\u5206\u5e03\u63d0\u4f9b\u53d8\u5316\u6a21\u5757\u7684\u5148\u9a8c\u77e5\u8bc6\u3002\u8be5\u65b9\u6cd5\u63d0\u4f9b\u4e86\u4e00\u79cd\u9886\u57df\u81ea\u9002\u5e94\u7684\u7aef\u5230\u7aef\u7684\u6846\u67b6\uff0c\u5728\u8fd9\u4e2a\u6846\u67b6\u4e2d\uff0c\u5982\u679c\u6709\u989d\u5916\u7684\u5173\u4e8e\u8054\u5408\u5206\u5e03\u53d8\u5316\u7684\u77e5\u8bc6\uff0c\u8fd9\u4e9b\u77e5\u8bc6\u5c31\u80fd\u88ab\u76f4\u63a5\u7eb3\u5165\u6846\u67b6\u4e2d\u4f7f\u7528\u4ece\u800c\u6539\u8fdb\u56fe\u7684\u8868\u793a\u3002\u6211\u4eec\u5728\u672c\u6587\u4e2d\u8ba8\u8bba\u4e86\u5982\u4f55\u5728\u6240\u63d0\u51fa\u7684\u6846\u67b6\u4e2d\u8868\u8ff0\u57fa\u4e8e\u56e0\u679c\u7684\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u3002\u5728\u4eff\u771f\u548c\u771f\u5b9e\u6570\u636e\u4e2d\u7684\u5b9e\u9a8c\u7ed3\u679c\u4e5f\u8bc1\u660e\u4e86\u6240\u63d0\u51fa\u7684\u9886\u57df\u81ea\u9002\u5e94\u6846\u67b6\u7684\u6709\u6548\u6027\u3002\u672c\u6587\u4e2d\u6d89\u53ca\u7684\u4ee3\u7801\u53ef\u4ee5\u5728  https:\/\/github.com\/mgong2\/DA_Infer \u83b7\u53d6\u3002<\/span><\/section>\n

<\/span><\/section>\n
\u8bd1\u8005<\/strong>\uff1a\u8881\u857e<\/span><\/section>\n
\u9f9a\u9e64\u626c\u7b80\u8bc4\uff1a<\/span><\/strong>\u8fd9\u662f\u89e3\u51b3\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u7684\u4e00\u79cd\u65b0\u601d\u8def\uff0c\u5176\u4e2d\u4e00\u4e2a\u5173\u952e\u5047\u8bbe\u662f\u4e0d\u540c\u9886\u57df\u6570\u636e\u7684\u90e8\u5206\u662f\u67d0\u4e2a mother \u5206\u5e03 i.i.d \u91c7\u6837\u5f97\u5230\u7684\u3002<\/span><\/section>\n

<\/section>\n

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

<\/section>\n

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

2. \u8fd1\u671f\u793e\u533a\u6d3b\u52a8<\/span><\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n
 <\/span>
<\/section>\n
2020\u5e7411\u670822\u65e5\uff0c\u56e0\u679c\u79d1\u5b66\u4e0eCausal AI\u8bfb\u4e66\u4f1a\u7b2c\u4e5d\u671f\u2014\u2014\u201c\u56e0\u679c\u63a8\u7406\u548c\u8fc1\u79fb\u5b66\u4e60\u201d\u5982\u671f\u8fdb\u884c\uff0c\u5bab\u660e\u660e\u3001\u90ed\u5bb6\u8d24\u3001\u4e01\u6668\u709c\u4f5c\u4e86\u7cbe\u5f69\u5206\u4eab\u3002<\/span><\/section>\n

<\/section>\n
\u4e3b\u8bb2\u4eba\u4ecb\u7ecd\uff1a<\/span><\/strong><\/section>\n
\u5bab\u660e\u660e\uff0c\u58a8\u5c14\u672c\u5927\u5b66\u8bb2\u5e08\uff0c\u7814\u7a76\u65b9\u5411\u4e3a\u56e0\u679c\u63a8\u65ad\uff0c\u57fa\u4e8e\u56e0\u679c\u7684\u673a\u5668\u5b66\u4e60\uff0c\u8fc1\u79fb\u5b66\u4e60\uff0c\u8ba1\u7b97\u673a\u89c6\u89c9\u3002<\/span><\/section>\n
\u90ed\u5bb6\u8d24\uff0c \u6089\u5c3c\u5927\u5b66\u5728\u8bfb\u535a\u58eb\uff0c\u7814\u7a76\u65b9\u5411\u4e3a\u6df1\u5ea6\u8fc1\u79fb\u5b66\u4e60\uff0c\u5f3a\u5316\u5b66\u4e60\u3002<\/span><\/section>\n
\u4e01\u6668\u709c\uff0c\u6089\u5c3c\u5927\u5b66\u5728\u8bfb\u535a\u58eb\uff0c\u7814\u7a76\u65b9\u5411\u4e3a\u56e0\u679c\u53d1\u73b0\uff0c\u8ba1\u7b97\u673a\u89c6\u89c9\u3002<\/span><\/section>\n

<\/span><\/section>\n
\u6f14\u8bb2\u5185\u5bb9\u7b80\u4ecb\uff1a<\/span><\/strong><\/section>\n
\u5927\u6570\u636e\u7684\u51fa\u73b0\u4f7f\u5f97\u8bb8\u591a\u5b66\u79d1\u5728\u5b66\u4e60\u548c\u9884\u6d4b\u65b9\u9762\u53d6\u5f97\u4e86\u9769\u547d\u6027\u7684\u6210\u529f\u3002\u4f46\u662f\uff0c\u5f53\u524d\u7684\u673a\u5668\u5b66\u4e60\u6a21\u578b\u7684\u4e00\u4e2a\u4e3b\u8981\u7f3a\u70b9\u662f\u7f3a\u4e4f\u5bf9\u65b0\u9886\u57df\u7684\u9002\u5e94\u6027\u548c\u6cdb\u5316\u80fd\u529b\u3002\u4e5f\u5c31\u662f\u8bf4\uff0c\u6807\u51c6\u7684\u76d1\u7763\u5b66\u4e60\u6a21\u578b\u5728\u6570\u636e\u5206\u5e03\u53d1\u751f\u53d8\u5316\u65f6\u9884\u6d4b\u6027\u80fd\u4f1a\u663e\u8457\u4e0b\u964d\u3002\u5728\u672c\u6b21\u8bfb\u4e66\u4f1a\uff0c\u6211\u4eec\u5c06\u91cd\u70b9\u8bb2\u8ff0\u5982\u679c\u5229\u7528\u56e0\u679c\u6a21\u578b\u7406\u89e3\u548c\u5efa\u6a21\u4e0d\u540c\u9886\u57df\u7684\u5206\u5e03\u53d8\u5316\u3002\u56e0\u4e3a\u56e0\u679c\u7cfb\u7edf\u7684\u72ec\u7acb\u6a21\u5757\u6027\u8d28\uff0c\u6211\u4eec\u53ef\u4ee5\u5c06\u590d\u6742\u7684\u5206\u5e03\u5206\u89e3\u6210\u5c0f\u7684\u6a21\u5757\uff0c\u53d1\u6398\u5206\u5e03\u5728\u4e0d\u540c\u9886\u57df\u7684\u4e0d\u53d8\u6027\u548c\u53d8\u5316\u6027\uff0c\u4ece\u800c\u5f00\u53d1\u51fa\u5177\u6709\u9886\u57df\u81ea\u9002\u5e94\u80fd\u529b\u7684\u9ad8\u6548\u8fc1\u79fb\u5b66\u4e60\u7b97\u6cd5\u3002\u4ee5\u4e0b\u662f\u672c\u6b21\u8bfb\u4e66\u4f1a\u7684\u5927\u7eb2\uff1a<\/span><\/section>\n

<\/span><\/section>\n
1\uff09\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u7684\u5b9a\u4e49<\/span><\/section>\n
2\uff09\u56e0\u679c\u6a21\u578b\u7684\u72ec\u7acb\u6a21\u5757\u6027\u8d28<\/span><\/section>\n
3\uff09\u56e0\u679c\u89c6\u89d2\u4e0b\u7684\u9886\u57df\u81ea\u9002\u5e94<\/span><\/section>\n
4\uff09\u4e0d\u540c\u9886\u57df\u81ea\u9002\u5e94\u8bbe\u5b9a\u4e0b\u7684\u7b97\u6cd5\u8bbe\u8ba1<\/span><\/section>\n
5\uff09\u591a\u9886\u57df\u7684\u56e0\u679c\u6a21\u578b\u5b66<\/span><\/section>\n

<\/span><\/section>\n
\u8bfb\u4e66\u4f1a\u7cbe\u5f69\u95ee\u7b54:<\/span><\/strong>
<\/span><\/section>\n

<\/section>\n
\u95ee\uff1a\u5171\u4eab\u53c2\u6570\u662f\u4ec0\u4e48\uff1f<\/span><\/strong><\/span><\/section>\n
\u7b54\uff1a<\/span><\/strong>\u6bcf\u4e2acausal module\uff0c\u53c2\u6570\u662f\u7531\u4e0d\u540c\u90e8\u5206\u53d8\u6210\u3002\u8fd9\u4e9b\u5c31\u662f\u540c\u6b65\u53d8\u5316\u7684\u53c2\u6570\uff0c\u53ef\u4ee5\u770b\u6210\u662fconfounder\u3002\u8fd9\u79cd\u53d8\u91cf\u53eb\u505a\u4e8c\u5c42\u53d8\u91cf\uff0c\u53eb\u5171\u540c\u5f71\u54cd\u7684\u56e0\u5b50\u3002\u516c\u53f8\u6587\u5316\u6216\u8005\u5236\u5ea6\u5bf9\u4e2a\u4f53\u7684\u5f71\u54cd\u3002C\u53ef\u4ee5\u662f\u4f01\u4e1a\u6587\u5316\uff0c\u6bcf\u4e2a\u4eba\u91c7\u6837\u90fd\u91c7V1\uff0cV2\uff0cV3\uff0cV4\u7684\u56db\u4e2a\u53d8\u91cf\uff0c\u6bd4\u5982\u5de5\u8d44\u591a\u5c11\uff0c\u52a0\u73ed\u65f6\u95f4\u7b49\u7b49\u3002\u5bf9\u4e8e\u540c\u4e00\u4e2a\u516c\u53f8\u7684\u4e0d\u540c\u4eba\u90fd\u662f\u4e00\u6837\u7684\uff0c\u53ea\u662f\u5f71\u54cd\u7684\u4e0d\u540c\u56e0\u7d20\u3002<\/span><\/section>\n

<\/section>\n
\u95ee\uff1aWi-Fi\u548caugmented\u7684\u771f\u5b9e\u5e94\u7528\u662f\u4ec0\u4e48\uff1f<\/span><\/strong>
<\/span><\/section>\n
\u7b54\uff1a<\/span><\/strong>Wifi\u53d8\u91cf\u4e4b\u95f4\u7684\u5173\u7cfb\u505aadaptation domain\u3002\u8fd9\u4e2a\u5e94\u7528\u672c\u8eab\u5c31\u662f\u505awifi\u5b9a\u4f4d\u7684\u65f6\u5019\uff0c\u77e5\u9053\u5177\u4f53\u4f4d\u7f6e\uff0c\u8fd9\u6837\u53ef\u4ee5\u641c\u5230\u4e00\u7cfb\u5217wifi\u4fe1\u53f7\uff0c\u901a\u8fc7\u6bcf\u4e2a\u8def\u7531\u5668\u4fe1\u53f7\u5f3a\u5f31\u6765\u5224\u65ad\u8ddd\u79bb\u3002\u65e9\u4e0a\u3001\u665a\u4e0a\u548c\u4e2d\u5348\u62cd\u51fa\u6765\u7684\u4fe1\u53f7\u968f\u7740\u65f6\u95f4\u53d8\u5316\uff0c\u4eca\u5929\u548c\u660e\u5929\u4e0d\u4e00\u6837\u3002\u8fd9\u4e2a\u5e94\u7528\u5c31\u662f\u7528\u4e00\u4e2amodel\u53ef\u4ee5\u7a7f\u5230\u5404\u79cd\u5404\u6837\u65f6\u95f4\u6bb5\u91cc\uff0c\u4e0d\u7528\u91c7\u96c6\u90a3\u4e48\u591a\u6570\u636e\u3002<\/span><\/section>\n

<\/section>\n
\u56e0\u679c\u79d1\u5b66\u793e\u533a\u7b80\u4ecb<\/strong>\uff1a\u5b83\u662f\u7531\u96c6\u667a\u4ff1\u4e50\u90e8\u3001\u667a\u6e90\u793e\u533a\u5171\u540c\u63a8\u52a8\uff0c\u9762\u5411\u56e0\u679c\u79d1\u5b66\u9886\u57df\u7684\u5782\u76f4\u578b\u5b66\u672f\u8ba8\u8bba\u793e\u533a\uff0c\u76ee\u7684\u662f\u4fc3\u8fdb\u56e0\u679c\u79d1\u5b66\u4e13\u4e1a\u4eba\u58eb\u548c\u5174\u8da3\u7231\u597d\u8005\u4eec\u7684\u4ea4\u6d41\u548c\u5408\u4f5c\uff0c\u63a8\u8fdb\u56e0\u679c\u79d1\u5b66\u5b66\u672f\u3001\u4ea7\u4e1a\u751f\u6001\u7684\u5efa\u8bbe\u548c\u843d\u5730\uff0c\u5b55\u80b2\u65b0\u4e00\u4ee3\u56e0\u679c\u79d1\u5b66\u9886\u57df\u7684\u5b66\u672f\u4e13\u5bb6\u548c\u4ea7\u4e1a\u521b\u65b0\u8005\u3002
<\/section>\n
   <\/span><\/section>\n
\u56e0\u679c\u79d1\u5b66\u793e\u533a\u6b22\u8fce\u60a8\u52a0\u5165\uff01<\/section>\n
 <\/span><\/section>\n
\u56e0\u679c\u79d1\u5b66\u793e\u533a\u613f\u666f<\/strong>\uff1a<\/em>\u56de\u7b54\u56e0\u679c\u95ee\u9898\u662f\u5404\u4e2a\u9886\u57df\u8feb\u5207\u7684\u9700\u6c42\uff0c\u5f53\u524d\u8bb8\u591a\u4e0d\u540c\u9886\u57df\uff08\u4f8b\u5982 AI \u548c\u7edf\u8ba1\u5b66\uff09<\/span>\u90fd\u5728\u4f7f\u7528\u56e0\u679c\u63a8\u7406\uff0c\u4f46\u662f\u4ed6\u4eec\u6240\u4f7f\u7528\u7684\u8bed\u8a00\u548c\u6a21\u578b\u5404\u4e0d\u76f8\u540c\uff0c\u5bfc\u81f4\u8fd9\u4e9b\u9886\u57df\u79d1\u5b66\u5bb6\u4e4b\u95f4\u6c9f\u901a\u4ea4\u6d41\u56f0\u96be\u3002\u56e0\u6b64\u6211\u4eec\u5e0c\u671b\u6784\u5efa\u4e00\u4e2a\u793e\u533a\uff0c\u901a\u8fc7\u7ec4\u7ec7\u5927\u91cf\u5b66\u672f\u6d3b\u52a8\uff0c\u4f7f\u5f97\u79d1\u7814\u4eba\u5458\u80fd\u591f\u638c\u63e1\u7edf\u8ba1\u5b66\u7684\u6838\u5fc3\u601d\u60f3\uff0c\u719f\u7ec3\u4f7f\u7528\u5f53\u524d AI \u5404\u79cd\u6280\u672f\uff08\u4f8b\u5982 Pytorch\/Pyro \u642d\u5efa\u6df1\u5ea6\u6982\u7387\u6a21\u578b\uff09<\/span>\uff0c\u4fc3\u8fdb\u5404\u4e2a\u9886\u57df\u7684\u7814\u7a76\u8005\u4ea4\u6d41\u548c\u601d\u7ef4\u78b0\u649e\uff0c\u4ece\u800c\u8ba9\u5404\u4e2a\u9886\u57df\u7684\u56e0\u679c\u63a8\u7406\u6709\u7740\u5171\u540c\u7684\u8303\u5f0f\uff0c\u751a\u81f3\u662f\u5171\u540c\u7684\u5de5\u7a0b\u5b9e\u8df5\u6807\u51c6\uff0c\u63a8\u52a8\u521a\u521a\u6210\u578b\u7684\u56e0\u679c\u79d1\u5b66\u5feb\u901f\u5411\u524d\u53d1\u5c55\u3002<\/strong>\u5177\u5907\u56e0\u679c\u63a8\u7406\u80fd\u529b\u7684\u4eba\u7c7b\u7d27\u5bc6\u534f\u4f5c\u521b\u9020\u4e86\u5f3a\u5927\u7684\u6587\u660e\uff0c\u6211\u4eec\u5e0c\u671b\u5728\u672a\u6765\u793e\u4f1a\u4e2d\uff0c\u56e0\u679c\u63a8\u7406\u878d\u5165\u5230\u6bcf\u4e2a\u5b66\u79d1\uff0c\u5c24\u5176\u662f\u7d27\u5bc6\u7ed3\u5408\u548c\u63d0\u5347 AI \uff0c\u671f\u5f85\u65e0\u6570\u5177\u5907\u6500\u767b\u56e0\u679c\u4e4b\u68af\u80fd\u529b\u7684 Agents (Causal AI) \u548c\u4eba\u7c7b\u4e00\u8d77\u534f\u4f5c\uff0c\u5171\u5efa\u4e0b\u4e00\u4ee3\u7684\u4eba\u7c7b\u6587\u660e\uff01<\/strong><\/section>\n
 <\/section>\n
\u5982\u679c\u60a8\u6709\u9002\u5f53\u7684\u6570\u5b66\u57fa\u7840\u548c\u4eba\u5de5\u667a\u80fd\u7814\u7a76\u7ecf\u9a8c\uff0c\u65e2\u6709\u79d1\u5b66\u5bb6\u7684\u597d\u5947\u5fc3\u4e5f\u6709\u5de5\u7a0b\u5e08\u601d\u7ef4\uff0c\u5e0c\u671b\u53c2\u4e0e\u5230\u201d\u56e0\u679c\u9769\u547d\u201c\u4e2d\uff0c\u6559\u4f1a\u673a\u5668\u56e0\u679c\u601d\u7ef4\uff0c\u4e3a\u56e0\u679c\u79d1\u5b66\u4f5c\u51fa\u8d21\u732e\uff0c\u8bf7\u52a0\u5165\u6211\u4eec\u5fae\u4fe1\u7fa4\uff1a\u626b\u63cf\u4e0b\u9762\u793e\u533a\u5c0f\u52a9\u624b\u4e8c\u7ef4\u7801\u52a0\u5165\uff08\u8bf7\u5907\u6ce8\u201c\u56e0\u679c\u79d1\u5b66\u201d\uff09\ud83d\udc47<\/span><\/strong><\/section>\n

<\/section>\n
        <\/section>\n

<\/section>\n
\u9605\u8bfb\u5f80\u671f\u300a\u56e0\u679c\u79d1\u5b66\u5468\u520a\u300b\uff0c\u8bf7\u70b9\u51fb\u4e0b\u9762\u94fe\u63a5\uff1a<\/span><\/strong><\/section>\n

<\/section>\n
\u300a\u56e0\u679c\u79d1\u5b66\u5468\u520a\u300b\u7b2c\u4e00\u671f\uff1a\u56e0\u679c\u793e\u533a\u8bda\u9080\u52a0\u5165\uff0c\u6253\u9020\u56e0\u679c\u63a8\u7406\u5171\u540c\u8303\u5f0f<\/span><\/a><\/section>\n
\u300a\u56e0\u679c\u79d1\u5b66\u5468\u520a\u300b\u7b2c\u4e8c\u671f\uff1a\u5982\u4f55\u89e3\u51b3\u6df7\u6dc6\u504f\u5dee\uff1f<\/span><\/a><\/section>\n
\u300a\u56e0\u679c\u79d1\u5b66\u5468\u520a\u300b\u7b2c\u4e09\u671f\uff1a\u56e0\u679c\u52a9\u529b Stable Learning<\/a><\/section>\n
\u300a\u56e0\u679c\u79d1\u5b66\u5468\u520a\u300b\u7b2c\u56db\u671f\uff1a\u56e0\u679c\u8d4b\u80fd\u63a8\u8350\u7cfb\u7edf<\/span><\/a><\/section>\n
\u300a\u56e0\u679c\u79d1\u5b66\u5468\u520a\u300b\u7b2c\u4e94\u671f\uff1aOOD\u6cdb\u5316<\/span><\/a>
<\/section>\n


<\/p>\n


<\/p>\n

\n
\n
\n
\n

\u7cfb\u5217\u8bfb\u4e66\u4f1a\u62a5\u540d<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n


<\/p>\n

\u5927\u6570\u636e\u65f6\u4ee3\u7684\u4e0b\u4e00\u573a\u53d8\u9769\u2014\u2014\u56e0\u679c\u9769\u547d\u6b63\u5728\u915d\u917f\u4e4b\u4e2d\uff0c\u901a\u8fc7\u878d\u5408\u56e0\u679c\u63a8\u7406\u548c\u673a\u5668\u5b66\u4e60\u800c\u6784\u5efa\u51fa\u6765\u7684Causal AI\u7cfb\u7edf\uff0c\u6709\u671b\u5960\u5b9a\u5f3a\u4eba\u5de5\u667a\u80fd\u7684\u57fa\u77f3\u3002<\/span><\/p>\n


<\/p>\n

\u96c6\u667a\u4ff1\u4e50\u90e8\u8054\u5408\u5317\u4eac\u667a\u6e90\u4eba\u5de5\u667a\u80fd\u7814\u7a76\u9662\uff0c\u9080\u8bf7\u4e86\u4e00\u6279\u5bf9\u56e0\u679c\u79d1\u5b66\u4e0eCausal AI\u611f\u5174\u8da3\u7684\u7814\u7a76\u8005\uff0c\u5f00\u5c55\u4e3a\u671f2-3\u4e2a\u6708\u7684\u7cfb\u5217\u7ebf\u4e0a\u8bfb\u4e66\u4f1a\uff0c\u7814\u8bfb\u7ecf\u5178\u548c\u524d\u6cbf\u8bba\u6587\uff0c\u5e76\u5c1d\u8bd5\u96c6\u4f53\u64b0\u5199\u4e00\u90e8\u4e66\u7c4d\u3002\u5982\u679c\u4f60\u4e5f\u4ece\u4e8b\u76f8\u5173\u7684\u7814\u7a76\u3001\u5e94\u7528\u5de5\u4f5c\uff0c\u6b22\u8fce\u62a5\u540d\uff0c\u53c2\u4e0e\u8bfb\u4e66\u4f1a\u7684\u8ba8\u8bba\uff01<\/span><\/p>\n

 <\/span><\/p>\n

\u65f6\u95f4\uff1a<\/span><\/strong>9\u670820\u65e5\u8d77\uff0c\u6bcf\u5468\u65e5\u665a19:00-21:00\uff0c\u6301\u7eed\u7ea62-3\u4e2a\u6708<\/span><\/section>\n
\u6a21\u5f0f\uff1a<\/span><\/strong>\u7ebf\u4e0a\u95ed\u95e8\u8bfb\u4e66\u4f1a\uff1b\u6536\u8d39-\u9000\u6b3e\u7684\u4fdd\u8bc1\u91d1\u6a21\u5f0f\uff1b\u8bfb\u4e66\u4f1a\u6210\u5458\u8ba4\u9886\u89e3\u8bfb\u8bba\u6587<\/span><\/section>\n
\u8d39\u7528\uff1a<\/span><\/strong>299\/\u4eba<\/span><\/section>\n

 <\/span><\/p>\n

\u4e86\u89e3\u8bfb\u4e66\u4f1a\u5177\u4f53\u89c4\u5219\u3001\u62a5\u540d\u8bfb\u4e66\u4f1a\u8bf7\u70b9\u51fb\u4e0b\u65b9\u6587\u7ae0\uff1a<\/span><\/p>\n

\u56e0\u679c\u79d1\u5b66\u4e0e Causal AI \u7cfb\u5217\u8bfb\u4e66\u4f1a | \u4f17\u5305\u51fa\u4e66<\/a><\/section>\n


<\/p>\n

\u76ee\u524d\u8bfb\u4e66\u4f1a\u5df2\u7ecf\u6709\u8d85\u8fc7130\u4f59\u4eba\u7684\u6d77\u5185\u5916\u9ad8\u6821\u79d1\u7814\u9662\u6240\u7684\u4e00\u7ebf\u79d1\u7814\u5de5\u4f5c\u8005\u4ee5\u53ca\u4e92\u8054\u7f51\u4e00\u7ebf\u4ece\u4e1a\u4eba\u5458\u53c2\u4e0e\uff0c\u5982\u679c\u4f60\u4e5f\u5bf9\u8fd9\u4e2a\u4e3b\u9898\u611f\u5174\u8da3\uff0c\u5c31\u5feb\u52a0\u5165\u6211\u4eec\u5427\uff01<\/span><\/p>\n

              <\/span><\/p>\n

\u56fe\u6ce8\uff1a\u9488\u5bf9\u8bfb\u4e66\u4f1a\u7684\u4e3b\u9898\uff0c\u7531\u53d1\u8d77\u4eba\u9f9a\u9e64\u626c\u8bbe\u7f6e\u597d\u4e86\u5185\u5bb9\u6846\u67b6\uff0c\u6bcf\u4e2a\u4e3b\u9898\u4e0b\u6709\u4e00\u4e2a\u8d1f\u8d23\u4eba\u6765\u8d1f\u8d23\u7ef4\u62a4\u7ec4\u7ec7\u76f8\u5173\u5185\u5bb9\uff0c\u76ee\u524d\u5df2\u7ecf\u5b9a\u597d\u7684\u5982\u56fe\u6240\u793a\uff0c\u6b22\u8fce\u5bf9\u4e3b\u9898\u611f\u5174\u8da3\u7684\u8054\u7cfb\u76f8\u5173\u8d1f\u8d23\u4eba\uff0c\u4ee5\u53ca\u6765\u8ba4\u9886\u76f8\u5173\u4e3b\u9898\u3002<\/em><\/span><\/p>\n<\/p><\/div>\n","protected":false},"excerpt":{"rendered":"

\u4e3a\u4e86\u5e2e\u52a9\u5927\u5bb6\u66f4\u597d\u5730\u4e86\u89e3\u56e0\u679c\u79d1\u5b66\u7684\u6700\u65b0\u79d1\u7814\u8fdb\u5c55\u548c\u8d44\u8baf\uff0c\u6211\u4eec\u56e0\u679c\u79d1\u5b66\u793e\u533a\u56e2\u961f\u672c\u5468\u6574\u7406\u4e86\u7b2c6\u671f\u300a\u56e0\u679c\u79d1\u5b66\u5468\u520a\u300b\uff0c\u63a8\u9001\u8fd1\u671f\u56e0\u679c\u79d1\u5b66\u503c\u5f97\u5173\u6ce8\u7684\u8bba\u6587\u548c\u8d44\u8baf\u4fe1\u606f\u3002\u672c\u671f\u7684\u4e3b\u9898\u662f\u201d\u9886\u57df\u81ea\u9002\u5e94\u201c\u3002 \u672c\u671f\u4f5c\u8005\uff1a\u5218\u73c8\u9e9f \u8bb8\u96c4\u9510 \u8d75\u6c5f\u6770 \u9648\u6657\u66e6 \u8881\u857e \u65b9\u6587\u6bc5 \u6768\u4e8c\u8336 \u9f9a\u9e64\u626c \u5bab\u660e\u660e \u5f53\u524d\u673a\u5668\u5b66\u4e60\u7814\u7a76\u7684\u4e00\u4e2a\u91cd\u70b9\uff0c\u662f\u7b97\u6cd5\u7684\u9c81\u68d2\u6027…<\/p>\n","protected":false},"author":1,"featured_media":22753,"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\/22756"}],"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"}],"author":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=22756"}],"version-history":[{"count":0,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/22756\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/media\/22753"}],"wp:attachment":[{"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=22756"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=22756"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=22756"},{"taxonomy":"special","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fspecial&post=22756"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}