{"id":24389,"date":"2021-03-10T19:31:56","date_gmt":"2021-03-10T11:31:56","guid":{"rendered":"https:\/\/swarma.org\/?p=24389"},"modified":"2021-03-10T19:31:56","modified_gmt":"2021-03-10T11:31:56","slug":"%e5%89%8d%e6%b2%bf%e7%bb%bc%e8%bf%b0%ef%bc%9a%e5%9b%a0%e6%9e%9c%e6%8e%a8%e6%96%ad%e4%b8%8e%e5%9b%a0%e6%9e%9c%e6%80%a7%e5%ad%a6%e4%b9%a0%e7%a0%94%e7%a9%b6%e8%bf%9b%e5%b1%95","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=24389","title":{"rendered":"\u524d\u6cbf\u7efc\u8ff0\uff1a\u56e0\u679c\u63a8\u65ad\u4e0e\u56e0\u679c\u6027\u5b66\u4e60\u7814\u7a76\u8fdb\u5c55"},"content":{"rendered":"
<\/p>\n

\u5bfc\u8bed<\/p>\n

\u672c\u6587\u4e3b\u8981\u4ece\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\u548c\u56e0\u679c\u6027\u5b66\u4e60\u8fd9\u4e24\u4e2a\u65b9\u9762\u4ecb\u7ecd\u548c\u5206\u6790\u4e86\u76ee\u524d\u56fd\u5185\u5916\u76f8\u5173\u9886\u57df\u7684\u7814\u7a76\u8fdb\u5c55\uff0c\u63a2\u8ba8\u4e86\u8fd9\u4e24\u65b9\u9762\u7814\u7a76\u7684\u53d1\u5c55\u65b9\u5411\u3002<\/p>\n

\u96c6\u667a\u4ff1\u4e50\u90e8\u8054\u5408\u667a\u6e90\u793e\u533a\uff0c\u7ec4\u7ec7\u56e0\u679c\u79d1\u5b66\u7cfb\u5217\u8bfb\u4e66\u4f1a\u7b2c\u4e8c\u5b63<\/a>\uff0c\u4ece\u57fa\u7840\u548c\u5b9e\u64cd\u89d2\u5ea6\u51fa\u53d1\uff0c\u7cbe\u8bfb\u4e24\u672c\u56e0\u679c\u79d1\u5b66\u65b9\u5411\u7684\u5165\u95e8\u6559\u6750\u3002\u672c\u6587\u4f5c\u8005\u8521\u745e\u521d\u8001\u5e08\u56e2\u961f\u4e5f\u4f1a\u5728\u8bfb\u4e66\u4f1a\u4e0a\u5c55\u5f00\u5206\u4eab\uff084\u6708\u5e95\uff09\uff0c\u8bfb\u4e66\u4f1a\u8be6\u60c5\u89c1\u6587\u672b\u3002
\u8521\u745e\u521d\u3001\u9648\u8587\u3001\u90dd\u5fd7\u5cf0 | \u4f5c\u8005
\u4e2d\u56fd\u4eba\u5de5\u667a\u80fd\u5b66\u4f1a | \u6765\u6e90<\/p>\n

 <\/p>\n

0 \u5f15\u8a00<\/p>\n

\u56e0\u679c\u5173\u7cfb\u4e00\u76f4\u662f\u4eba\u7c7b\u8ba4\u8bc6\u4e16\u754c\u7684\u57fa\u672c\u65b9\u5f0f\u548c\u73b0\u4ee3\u79d1\u5b66\u7684\u57fa\u77f3\u3002\u7231\u56e0\u65af\u5766\u5c31\u66fe\u6307\u51fa\uff0c\u897f\u65b9\u79d1\u5b66\u7684\u53d1\u5c55\u662f\u4ee5\u5e0c\u814a\u54f2\u5b66\u5bb6\u53d1\u660e\u5f62\u5f0f\u903b\u8f91\u4f53\u7cfb\uff0c\u4ee5\u53ca\u901a\u8fc7\u7cfb\u7edf\u7684\u5b9e\u9a8c\u53d1\u73b0\u6709\u53ef\u80fd\u627e\u51fa\u56e0\u679c\u5173\u7cfb\u8fd9\u4e24\u4e2a\u4f1f\u5927\u7684\u6210\u5c31\u4e3a\u57fa\u7840\u3002\u4ece\u4e0e\u76f8\u5173\u5173\u7cfb\u5bf9\u6bd4\u7684\u89d2\u5ea6\u6765\u770b\uff0c \u56e0\u679c\u5173\u7cfb\u4e25\u683c\u533a\u5206\u4e86\u201c\u539f\u56e0\u201d\u53d8\u91cf\u548c\u201c\u7ed3\u679c\u201d \u53d8\u91cf\uff0c\u5728\u63ed\u793a\u4e8b\u7269\u53d1\u751f\u673a\u5236\u548c\u6307\u5bfc\u5e72\u9884\u884c\u4e3a\u7b49\u65b9 \u9762\u6709\u76f8\u5173\u5173\u7cfb\u4e0d\u80fd\u66ff\u4ee3\u7684\u91cd\u8981\u4f5c\u7528\u3002\u4ee5\u56fe1\u4e3a\u4f8b\uff0c\u5438\u70df\u3001\u9ec4\u7259\u90fd\u4e0e\u80ba\u764c\u5177\u6709\u8f83\u5f3a\u7684\u76f8\u5173\u5173\u7cfb\uff0c\u7136\u800c\u53ea\u6709\u5438\u70df\u624d\u662f\u80ba\u764c\u7684\u539f\u56e0\uff0c\u4e5f\u53ea\u6709\u6212\u70df\u624d\u80fd\u964d\u4f4e\u80ba\u764c\u7684\u53d1\u75c5\u6982\u7387\uff0c\u800c\u628a\u7259\u9f7f\u6d17\u767d\u5219\u4e0d\u80fd\u964d\u4f4e\u80ba\u764c\u7684\u53d1\u75c5\u6982\u7387\u3002\u63a2\u7d22\u548c\u63a8\u65ad\u4e8b\u7269\u95f4\u7684\u56e0\u679c\u5173\u7cfb\uff0c\u662f\u6570\u636e\u79d1\u5b66\u4e2d\u7684\u4e00\u4e2a\u6838\u5fc3\u95ee\u9898\uff0c\u6b63\u53d7\u5230\u56fd\u5185\u5916\u540c\u884c\u7684\u5e7f\u6cdb\u5173\u6ce8\u3002\u73b0\u6709\u56e0\u679c\u5173\u7cfb\u7684\u7814\u7a76\u96c6\u4e2d\u5728\u56e0\u679c\u63a8\u65ad\u53ca\u56e0\u679c\u6027\u5b66\u4e60\u4e24\u4e2a\u65b9\u9762\u3002\u56fe 1  \u5438\u70df\u4e0e\u9ec4\u7259\u3001\u80ba\u764c\u4e4b\u95f4\u7684\u56e0\u679c\u5173\u7cfb\u53d1\u73b0 
\u56e0\u679c\u63a8\u65ad\u7684\u76ee\u6807\u662f\u53d1\u73b0\u53d8\u91cf\/\u4e8b\u7269\u80cc\u540e\u7684\u56e0\u679c\u5173\u7cfb\u3002\u968f\u673a\u63a7\u5236\u5b9e\u9a8c\u662f\u53d1\u73b0\u56e0\u679c\u5173\u7cfb\u7684\u4f20\u7edf\u65b9\u6cd5\u3002\u7531\u4e8e\u5b9e\u9a8c\u6280\u672f\u5c40\u9650\u548c\u5b9e\u9a8c\u8017\u8d39\u4ee3\u4ef7\u5de8\u5927\u7b49\u539f\u56e0\uff0c\u8d8a\u6765\u8d8a\u591a\u7684\u56e0\u679c\u63a8\u65ad\u9886\u57df\u5b66\u8005\u5e0c\u671b\u901a\u8fc7\u89c2\u5bdf\u6570\u636e\u63a8\u65ad\u53d8\u91cf\u4e4b\u95f4\u7684\u56e0\u679c\u5173\u7cfb\uff0c\u5df2\u6210\u4e3a\u5f53\u524d\u56e0\u679c\u63a8\u65ad\u9886\u57df\u7684\u7814\u7a76\u70ed\u70b9\u3002\u5728\u57fa\u4e8e\u89c2\u5bdf\u6570\u636e\u7684\u56e0\u679c\u63a8\u65ad\u9886\u57df\u7814\u7a76\u65b9\u9762\u7684\u4ee3\u8868\u6027\u8fdb\u5c55\u5305\u62ec\u5728\u4e0a\u4e2a\u4e16\u7eaa90\u5e74\u4ee3\uff0c\u56fe\u7075\u5956\u5f97\u4e3bJudea Pearl\u6559\u6388\u3001\u5361\u5185\u57fa\u6885\u9686\u5927\u5b66Clark Glymour\u6559\u6388\u7b49\u5148\u9a71\u5171\u540c\u5efa\u7acb\u4e86\u57fa\u4e8e\u89c2\u5bdf\u6570\u636e\u56e0\u679c\u63a8\u65ad\u7684\u7406\u8bba\u57fa\u7840\u548c\u57fa\u4e8e\u7ea6\u675f\u7684\u65b9\u6cd5\uff0c\u4ee5\u53ca\u8fd110\u5e74Bernhard Sch\u00f6lkopf\u3001Kun Zhang\u3001Shohei Shimizu\u7b49\u5b66\u8005\u4e3a\u4ee3\u8868\u63d0\u51fa\u7684\u57fa\u4e8e\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5\u3002 
\u56e0\u679c\u6027\u5b66\u4e60\u5219\u4f53\u73b0\u4e86\u56e0\u679c\u63a8\u65ad\u5bf9\u4e8e\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u8bbe\u8ba1\u7684\u6307\u5bfc\u4f5c\u7528\u3002\u968f\u7740\u4eba\u5de5\u667a\u80fd\u7684\u53d1\u5c55\uff0c\u8d8a\u6765\u8d8a\u591a\u5b66\u8005\u5f00\u59cb\u8ba4\u8bc6\u5230\u56e0\u679c\u63a8\u65ad\u5bf9\u4e8e\u514b\u670d\u73b0\u6709\u4eba\u5de5\u667a\u80fd\u65b9\u6cd5\/\u6280\u672f\u5728\u62bd\u8c61\u3001\u63a8\u7406\u548c\u53ef\u89e3\u91ca\u6027\u7b49\u65b9\u9762\u7684\u4e0d\u8db3\u5177\u6709\u91cd\u8981\u610f\u4e49\u3002\u6b63\u5982\u56fe\u7075\u5956\u5f97\u5956\u8005Judea Pearl\u5728\u65b0\u4f5c\u300aThe Book of Why\u300b\u4e00\u4e66\u4e2d\u63d0\u51fa\u7684 \u201c\u56e0\u679c\u5173\u7cfb\u4e4b\u68af\u201d\uff0c\u4ed6\u628a\u56e0\u679c\u63a8\u65ad\u5206\u6210\u4e09\u4e2a\u5c42\u9762\uff0c\u7b2c\u4e00\u5c42\u662f\u201c\u5173\u8054\u201d\uff1b\u7b2c\u4e8c\u5c42\u662f\u201c\u5e72\u9884\u201d\uff1b\u7b2c\u4e09\u5c42\u662f\u201c\u53cd\u4e8b\u5b9e\u63a8\u7406\u201d\u3002\u4ed6\u7279\u522b\u6307\u51fa\uff0c\u6211\u4eec\u5f53\u524d\u7684\u673a \u5668\u5b66\u4e60\u9886\u57df\u7684\u7814\u7a76\u53ea\u5904\u4e8e\u7b2c\u4e00\u5c42\uff0c\u53ea\u662f\u201c\u5f31\u4eba\u5de5\u667a\u80fd\u201d\uff0c\u8981\u5b9e\u73b0\u201c\u5f3a\u4eba\u5de5\u667a\u80fd\u201d\u8fd8\u9700\u8981\u5e72\u9884\u548c\u53cd\u4e8b\u5b9e\u63a8\u7406\u3002\u5728Kun Zhang\u7b49\u5b66\u8005\u53d1\u8868\u7684\u300aLearning causality and causality-related learning: some recent progress\u300b[1]\u7efc\u8ff0\u4e2d\uff0c\u5bf9\u57fa\u4e8e\u56e0\u679c\u601d\u60f3\u7684\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u8fdb\u884c\u4e86\u603b\u7ed3\uff0c\u63d0\u51fa\u4e86\u56e0\u679c\u6027\u5b66\u4e60\u8fd9\u4e00\u6982\u5ff5\u3002 
\u56e0\u679c\u63a8\u65ad\u3001\u56e0\u679c\u6027\u5b66\u4e60\u53ca\u4e00\u4e9b\u76f8\u5173\u6982\u5ff5\u7684\u5173\u7cfb\u5982\u56fe2\u6240\u793a\u3002\u56e0\u679c\u63a8\u65ad\u7406\u8bba\u548c\u65b9\u6cd5\u4e3a\u56e0\u679c\u6027\u5b66\u4e60\u63d0\u4f9b\u4e86\u91cd\u8981\u7684\u7406\u8bba\u57fa\u7840\u548c\u601d\u60f3\u7684\u6765\u6e90\u3002\u4e0b\u9762\u5206\u522b\u5bf9\u57fa\u4e8e\u89c2\u5bdf\u6570\u636e\u7684\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\u548c\u56e0\u679c\u6027\u5b66\u4e60\u65b9\u6cd5\u8fd9\u4e24\u4e2a\u65b9\u9762\u7684\u7814\u7a76\u8fdb\u5c55\u8fdb\u884c\u91cd\u70b9\u63a2\u8ba8\u3002\u56fe 2  \u672c\u6587\u7684\u7814\u7a76\u5185\u5bb9
 <\/p>\n


1 \u56e0\u679c\u63a8\u65ad<\/p>\n

\u8fd130\u5e74\u6765\uff0c\u56e0\u679c\u63a8\u65ad\u7684\u7814\u7a76\u4e00\u76f4\u5907\u53d7\u5b66\u8005\u4eec\u5173\u6ce8\u3002\u9488\u5bf9\u89c2\u5bdf\u6570\u636e\u7279\u6027\u7684\u4e0d\u540c\uff0c\u57fa\u4e8e\u89c2\u5bdf\u6570\u636e\u7684\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\u53ef\u4ee5\u5206\u4e3a\u57fa\u4e8e\u65f6\u5e8f\u89c2\u5bdf\u6570\u636e\u7684\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\u548c\u57fa\u4e8e\u975e\u65f6\u5e8f\u89c2\u5bdf\u6570\u636e\u7684\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\u3002\u867d\u7136\u65f6\u5e8f\u89c2\u5bdf\u6570\u636e\u4e2d\u65f6\u95f4\u7ef4\u5ea6\u8574\u542b\u4e86\u201c\u56e0-\u679c\u201d\u65b9\u5411\u7684\u91cd\u8981\u4fe1\u606f\u2014\u2014\u201c\u679c\u201d\u5728\u65f6\u95f4\u4e0a\u4e0d\u80fd\u53d1\u751f\u5728\u201c\u56e0\u201d\u7684\u524d\u9762\uff0c\u4f46\u662f\u65f6\u5e8f\u6570\u636e\u9700\u8981\u83b7\u53d6\u4e00\u4e2a\u5bf9\u8c61\u5728\u4e0d\u540c\u65f6\u523b\u7684\u89c2\u5bdf\u503c\uff0c\u5bf9\u89c2\u5bdf\u624b\u6bb5\u5177\u6709\u8f83\u9ad8\u7684\u8981\u6c42\u3002\u540c\u65f6\uff0c\u90e8\u5206\u57fa\u4e8e\u975e\u65f6\u5e8f\u89c2\u5bdf\u6570\u636e\u7684\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\u53ef\u4ee5\u62d3\u5c55\u5230\u65f6\u5e8f\u89c2\u5bdf\u6570\u636e\u4e2d\uff0c\u6240\u4ee5\u672c\u6587\u4e3b\u8981\u63a2\u8ba8\u57fa\u4e8e\u975e\u65f6\u5e8f\u89c2\u5bdf\u6570\u636e\u7684\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\u3002
\u56e0\u679c\u63a8\u65ad\u7684\u4e3b\u8981\u76ee\u7684\u662f\u4ece\u89c2\u5bdf\u6570\u636e\u4e2d\u63a8\u65ad\u53d8\u91cf\u4e4b\u95f4\u7684\u56e0\u679c\u5173\u7cfb\u3002\u76ee\u524d\uff0c\u5b66\u8005\u4eec\u4e3b\u8981\u5173\u6ce8\u56e0\u679c\u5173\u7cfb\u65b9\u5411\u63a8\u65ad\u3001\u9ad8\u7ef4\u6570\u636e\u4e0a\u7684\u8bef\u53d1\u73b0\u7387\u63a7\u5236\u548c\u4e0d\u5b8c\u5168\u89c2\u5bdf\u6570\u636e\u4e0a\u7684\u9690\u53d8\u91cf\u68c0\u6d4b\u7b49\u95ee\u9898\u3002\u5176\u4e2d\uff0c\u56e0\u679c\u5173\u7cfb\u65b9\u5411\u63a8\u65ad\u4e3b\u8981\u9488\u5bf9\u4e24\u4e2a\u53d8\u91cf\u8fdb\u884c\u7814\u7a76\uff0c\u65e8\u5728\u63a8\u65ad\u4e24\u4e2a\u53d8\u91cf\u4e4b\u95f4\u7684\u56e0\u679c\u5173\u7cfb\u65b9\u5411\uff0c\u5373\u8bc6\u522b\u54ea\u4e2a\u53d8\u91cf\u662f\u539f\u56e0\u53d8\u91cf\uff0c\u54ea\u4e2a\u53d8\u91cf\u662f\u7ed3\u679c\u53d8\u91cf\u3002\u9ad8\u7ef4\u6570\u636e\u4e0a\u7684\u56e0\u679c\u63a8\u65ad\u5219\u662f\u9488\u5bf9\u591a\u4e2a\u53d8\u91cf\u800c\u8a00\uff0c\u4e3b\u8981\u5173\u6ce8\u5982\u4f55\u964d\u4f4e\u56e0\u679c\u5173\u7cfb\u7ed3\u6784\u7684\u8bef\u53d1\u73b0\u7387\u8fd9\u4e00\u96be\u9898\u3002\u4e0d\u5b8c\u5168\u89c2\u5bdf\u6570\u636e\u4e0a\u7684\u9690\u53d8\u91cf\u68c0\u6d4b\u5219\u662f\u9488\u5bf9{X1,X2,..,Xn}\u4e0a\u7684\u89c2\u5bdf\u6570\u636e\uff0c\u68c0\u6d4b\u9690\u53d8\u91cf\u7684\u5b58\u5728\u6027\uff0c\u4ee5\u53ca\u63a8\u65ad\u9690\u53d8\u91cf\u4e0e\u89c2\u5bdf\u53d8\u91cf\u4e4b\u95f4\u7684\u56e0\u679c\u5173\u7cfb\u3002\u8fd9\u91cc\u7684\u9690\u53d8\u91cf\u662f\u6307\u672a\u80fd\u89c2\u5bdf\u5230\u6216\u65e0\u6cd5\u5ea6\u91cf\u7684\u53d8\u91cf\uff0c\u5373\u6570\u636e\u96c6\u4e2d\u6ca1\u6709\u5305\u542b\u7684\u53d8\u91cf\u3002<\/p>\n

\u901a\u5e38\u6765\u8bf4\uff0c\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\u53ef\u4ee5\u5206\u4e3a\u57fa\u4e8e\u7ea6\u675f\u7684\u65b9\u6cd5\u3001\u57fa\u4e8e\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5\u548c\u6df7\u5408\u578b\u65b9\u6cd5\u3002\u8fd9\u4e09\u7c7b\u65b9\u6cd5\u57fa\u672c\u4e0a\u56f4\u7ed5\u7740\u56e0\u679c\u5173\u7cfb\u65b9\u5411\u63a8\u65ad\u96be\u9898\u3001\u9ad8\u7ef4\u6570\u636e\u4e0a\u7684\u8bef\u53d1\u73b0\u7387\u63a7\u5236\u96be\u9898\u548c\u4e0d\u5b8c\u5168\u89c2\u5bdf\u6570\u636e\u4e0a\u7684\u9690\u53d8\u91cf\u68c0\u6d4b\u96be\u9898\u800c\u5c55\u5f00\uff0c\u5176\u76f8\u5173\u7684\u4e3b\u8981\u65b9\u6cd5\u5f52\u7eb3\u5982\u88681\u6240\u793a\u3002<\/p>\n

\u88681  \u56e0\u679c\u63a8\u65ad\u76f8\u5173\u65b9\u6cd5

1.1 \u57fa\u4e8e\u7ea6\u675f\u7684\u65b9\u6cd5
\u57fa\u4e8e\u7ea6\u675f\u7684\u65b9\u6cd5\u4e3b\u8981\u4ee5\u7f8e\u56fd\u5361\u5185\u57fa\u6885\u9686\u5927\u5b66Glymour\u6559\u6388\u548cSpirtes\u6559\u6388\u7684PC\uff08Peter-Clark\uff09\u7b97\u6cd5\uff0c\u4ee5\u53ca\u52a0\u5229\u798f\u5c3c\u4e9a\u5dde\u5927\u5b66\u6d1b\u6749\u77f6\u5206\u6821Pearl\u6559\u6388\u548cVerma\u6559\u6388\u7684IC\uff08Inductive Causation\uff09\u7b97\u6cd5\u4e3a\u4ee3\u8868\u3002\u8fd9\u4e24\u4e2a\u7b97\u6cd5\u7684\u57fa\u672c\u6d41\u7a0b\u4e3b\u8981\u6709\u4e24\u4e2a\u9636\u6bb5\uff0c\u9996\u5148\u5229\u7528\u57fa\u4e8e\u72ec\u7acb\u6027\u6216\u6761\u4ef6\u72ec\u7acb\u6027\u68c0\u9a8c\u7684\u76f8\u5173\u65b9\u6cd5\u5224\u65ad\u53d8\u91cf\u4e4b\u95f4\u7684\u72ec\u7acb\u6027\uff0c\u83b7\u5f97\u53d8\u91cf\u95f4\u7684\u56e0\u679c\u65e0\u5411\u56fe\uff1b\u7136\u540e\u518d\u5229\u7528V-\u7ed3\u6784\u548c\u5b9a\u5411\u89c4\u5219\u5bf9\u53d8\u91cf\u95f4\u7684\u65e0\u5411\u8fb9\u8fdb\u884c\u5b9a\u5411\u3002\u8fd9\u4e24\u4e2a\u7b97\u6cd5\u4e3b\u8981\u89e3\u51b3\u56e0\u679c\u5173\u7cfb\u65b9\u5411\u63a8\u65ad\u96be\u9898\u3002\u540e\u7eed\u6709\u4e0d\u5c11\u5b66\u8005\u5728\u6b64\u7b97\u6cd5\u4e0a \u8fdb\u884c\u4e86\u4e00\u4e9b\u62d3\u5c55\u548c\u6539\u8fdb\u3002
\u4e3a\u4e86\u964d\u4f4e\u9ad8\u7ef4\u6570\u636e\u4e0a\u7684\u8bef\u53d1\u73b0\u7387\uff0c\u5317\u4eac\u5927\u5b66\u803f\u76f4\u6559\u6388\u7b49\u63d0\u51fa\u4e86\u4e00\u79cd\u641c\u7d22\u5c40\u90e8\u7ed3\u6784\u7684\u5206\u89e3\u65b9\u6cd5\uff0c\u901a\u8fc7\u9012\u5f52\u65b9\u6cd5\u5c06\u56fe\u4e00\u5206\u4e3a\u4e8c\uff0c\u5b66\u4e60\u5c40\u90e8\u7ed3\u6784\uff0c\u5e76\u9010\u6b65\u81ea\u5e95\u5411\u4e0a\u6574\u5408\u6210\u5168\u5c40\u7ed3\u6784\u3002Tsamardinos\u7b49\u5219\u7ed3\u5408\u57fa\u4e8e\u7ea6\u675f\u7684\u65b9\u6cd5\u548c\u8d2a\u5a6a\u7b49\u4ef7\u7c7b\u641c\u7d22\u65b9\u6cd5\uff0c\u63d0 \u51fa\u4e86\u6700\u5927-\u6700\u5c0f\u722c\u5c71\u6cd5\uff08MMHC\uff09\u3002\u8fd9\u79cd\u65b9\u6cd5\u5148\u901a\u8fc7\u5c40\u90e8\u7ed3\u6784\u5b66\u4e60\u7b97\u6cd5\u2014\u2014\u6700\u5927-\u6700\u5c0f\u7236\u4eb2\u5b69\u5b50\uff08MMPC\uff09\u7b97\u6cd5\u5b66\u4e60\u56e0\u679c\u65e0\u5411\u56fe\uff0c\u7136\u540e\u7528\u8d2a\u5a6a\u8d1d\u53f6\u65af\u8bc4\u5206\u722c\u5c71\u641c\u7d22\u65b9\u6cd5\u5bf9\u65e0\u5411\u56fe\u8fdb\u884c\u5b9a\u5411\u3002
\u53e6\u4e00\u7c7b\u5b66\u8005\u4eec\u5173\u6ce8\u7684\u95ee\u9898\u662f\u4e0d\u5b8c\u5168\u89c2\u5bdf\u6570\u636e\u60c5\u51b5\u4e0b\u5b58\u5728\u9690\u53d8\u91cf\uff0c\u4e0d\u5b8c\u5168\u89c2\u5bdf\u6570\u636e\u5bfc\u81f4\u7684\u865a\u5047\u56e0\u679c\u5173\u7cfb\u95ee\u9898\u3002\u4f8b\u5982\uff0c\u56fe1\u7684\u4f8b\u5b50\u4e2d\uff0c\u5982\u679c \u53ea\u6709\u9ec4\u7259\u548c\u80ba\u764c\u7684\u76f8\u5173\u6570\u636e\uff0c\u5229\u7528\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\uff0c\u6211\u4eec\u5f80\u5f80\u4f1a\u53d1\u73b0\u4e8c\u8005\u4e4b\u95f4\u5b58\u5728\u56e0\u679c\u5173\u7cfb\u3002\u4f46\u662f\u5b9e\u9645\u4e0a\u8fd9\u79cd\u5173\u7cfb\u662f\u865a\u5047\u7684\uff0c\u672a\u89c2\u5bdf\u5230\u7684\u5438\u70df\u624d\u662f\u4e24\u8005\u7684\u5171\u540c\u539f\u56e0\uff0c\u9ec4\u7259\u548c\u80ba\u764c\u4e4b\u95f4\u5728\u5438\u70df\u6761\u4ef6\u4e0b\u662f\u4e92\u76f8\u72ec\u7acb\u7684\u3002\u5728\u73b0\u5b9e\u573a\u666f\u4e2d\u7ecf\u5e38\u51fa\u73b0\u8fd9\u79cd\u60c5\u51b5\uff0c\u8fd9\u65f6\u5bf9\u9690\u53d8\u91cf\u7684\u68c0\u6d4b\u5c31\u81f3\u5173\u91cd\u8981\u3002\u9488\u5bf9\u9690\u53d8\u91cf\u95ee\u9898\uff0cSpirtes\u6559\u6388\u7b49\u63d0\u51fa\u4e86FCI\uff08Fast Causal Inference\uff09\u7b97\u6cd5\uff0c\u540e\u7eed\u5b66\u8005\u4eec\u5bf9\u5176\u8fdb\u884c\u4e86\u62d3\u5c55\uff0c\u5982Colombo\u7b49\u7684RFCI\uff08Really Fast Causal Inference\uff09\u7b97\u6cd5\u3002\u8003\u8651\u5230\u5c0f\u6837\u672c\u7684\u60c5\u51b5\uff0cOgarrio \u7b49\u63d0\u51fa\u4e86GFCI\uff08Greedy Fast Causal Inference\uff09\u7b97\u6cd5\u3002\u9274\u4e8e\u7ebf\u6027\u56fe\u6a21\u578b\u8574\u542b\u7740\u591a\u79cd\u534f\u65b9\u5dee\u77e9\u9635\u5b50\u77e9\u9635\u7684\u6392\u5e8f\u7ea6\u675f\uff0cKummerfeld\u7b49\u5229\u7528\u8fd9\u4e9b\u6392\u5e8f\u7ea6\u675f\uff0c\u518d\u52a0\u4e0a\u6761\u4ef6\u72ec\u7acb\u6027\u68c0\u9a8c\uff0c\u63d0\u51fa\u4e86\u4e00\u79cdFTFC \uff08Find Two Factor Clusters\uff09\u7b97\u6cd5\uff0c\u7528\u4e8e\u8bc6\u522b\u9690\u53d8\u91cf\u6a21\u578b\u3002
1.2 \u57fa\u4e8e\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5
\u57fa\u4e8e\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5\u5219\u662f\u4ece\u6570\u636e\u4ea7\u751f\u7684\u56e0\u679c\u673a\u5236\u51fa\u53d1\uff0c\u63a2\u7d22\u5229\u7528\u56e0\u679c\u51fd\u6570\u6a21\u578b\u6765\u8bc6\u522b\u56e0\u679c\u65b9\u5411\u3002\u6b64\u7c7b\u65b9\u6cd5\u4e3b\u8981\u4ee5\u7ebf\u6027\u975e\u9ad8\u65af\u65e0\u73af\u6a21\u578b \uff08LiNGAM\uff09\u3001\u52a0\u6027\u566a\u58f0\u6a21\u578b\uff08ANM\uff09\uff0c\u540e\u975e\u7ebf\u6027\u6a21\u578b\uff08PNL\uff09\u548c\u4fe1\u606f\u51e0\u4f55\u65b9\u6cd5\uff08IGCI\uff09\u8fd9\u51e0\u7c7b\u6a21\u578b\u4e3a\u4ee3\u8868\u3002
\u7ebf\u6027\u65e0\u73af\u6a21\u578b\u662f\u4e00\u79cd\u8f83\u4e3a\u7ecf\u5178\u7684\u6a21\u578b\uff0c\u4e3b\u8981\u7528\u4e8e\u5206\u6790\u8fde\u7eed\u53d8\u91cf\u4e4b\u95f4\u7684\u56e0\u679c\u65b9\u5411\u4e0e\u56e0\u679c\u8fde\u63a5\u6743\u91cd\u3002\u5229\u7528\u6570\u636e\u7684\u975e\u9ad8\u65af\u6027\uff0cShimizu\u7b49\u4e8e2006\u5e74\u9996\u5148\u63d0\u51fa\u4e86LiNGAM\uff0c\u5e76\u7528\u72ec\u7acb\u6210\u5206\u5206\u6790\uff08ICA\uff09\u6c42\u89e3\uff0c\u6240\u4ee5\u53c8\u79f0\u4e3aICA-LiNGAM\u7b97\u6cd5\u3002\u4f46\u8be5\u6a21\u578b\u5177\u6709\u5c40\u90e8\u6536\u655b\u7684\u7f3a\u9677\uff0c\u4f7f\u5f97\u6c42\u89e3\u7ed3\u679c\u5f80\u5f80\u662f\u5c40\u90e8\u6700\u4f18\u89e3\uff0c\u800c\u4e0d\u662f\u5168\u5c40\u6700\u4f18\u89e3\u30022011\u5e74\uff0cShimizu \u7b49\u7d27\u63a5\u5730\u63d0\u51fa\u4e86DirectLiNGAM\uff08A Direct Method for a Linear Non-Gaussian SEM\uff09\u6846\u67b6\uff0c\u901a\u8fc7\u4e0d\u65ad\u5730\u8bc6\u522b\u5916\u751f\u53d8\u91cf\u8fdb\u800c\u4f30\u8ba1\u56e0\u679c\u6b21\u5e8f\u3002
\u4e0e\u7ebf\u6027\u6a21\u578b\u76f8\u6bd4\uff0c\u975e\u7ebf\u6027\u52a0\u566a\u6a21\u578b\u4e0d\u5177\u6709\u4f20\u9012\u6027\uff0c\u5373\u6bcf\u4e2a\u76f4\u63a5\u56e0\u679c\u5173\u7cfb\u9075\u5faa\u8be5\u6a21\u578b\uff0c\u4f46\u5374\u7701\u7565\u4e86\u4e2d\u95f4\u56e0\u679c\u53d8\u91cf\u3002\u56e0\u6b64\u6211\u4eec\u63d0\u51fa\u4e86\u4e00\u79cd\u7ea7\u8054\u975e\u7ebf\u6027\u52a0\u6027\u566a\u58f0\u6a21\u578b\uff08Cascade Nonlinear Additive Noise Models\uff09\u6765\u8868\u793a\u8fd9\u79cd\u56e0\u679c\u5173\u7cfb\uff0c\u5e76\u8fdb\u4e00\u6b65\u63d0\u51fa\u4e86\u4e00\u79cd\u5728\u53d8\u5206\u81ea\u52a8\u7f16\u7801\u5668\u6846\u67b6\u4e0b\u4ece\u6570\u636e\u4e2d\u4f30\u8ba1\u6a21\u578b\u7684\u65b9\u6cd5\u3002\u5b9e\u9a8c\u7ed3\u679c\u8868\u660e\uff0c\u6240\u63d0\u51fa\u7684\u6a21\u578b\u548c\u65b9\u6cd5\u6781\u5927\u5730\u6269\u5c55\u4e86\u57fa\u4e8e\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5\u5728\u975e\u7ebf\u6027\u60c5\u5f62\u4e2d\u7684\u9002\u7528\u6027\u3002
\u4e0a\u8ff0\u65b9\u6cd5\u4e3b\u8981\u9002\u7528\u4e8e\u8fde\u7eed\u578b\u6570\u636e\u3002\u76ee\u524d\u5c06\u8fde\u7eed\u7a7a\u95f4\u4e0a\u7684\u56e0\u679c\u65b9\u6cd5\u63a8\u5e7f\u5230\u79bb\u6563\u7a7a\u95f4\u4e0a\uff0c\u4ecd\u7136\u662f\u4e00\u4e2a\u5177\u6709\u6311\u6218\u7684\u95ee\u9898\u3002Peters\u7b49\u5c1d\u8bd5\u5c06\u52a0\u6027\u566a\u58f0\u6a21\u578b\u63a8\u5e7f\u5230\u79bb\u6563\u7684\u6570\u636e\u4e0a\uff0c\u7136\u800c\u5bf9\u4e8e\u7c7b\u522b\u578b\u53d8\u91cf\u6765\u8bf4\uff0c\u57fa\u4e8e\u52a0\u6027\u566a\u58f0\u6a21\u578b\u7684\u5047\u8bbe\u5f88\u96be\u88ab\u6ee1\u8db3\u3002\u6211\u4eec\u8bd5\u56fe\u627e\u5230\u4e00\u79cd\u66f4\u4e3a\u4e00\u822c\u7684\uff0c\u53ef\u9002\u7528\u4e8e\u66f4\u591a\u6570\u636e\u4e0a\u7684\u56e0\u679c\u673a\u5236\u3002\u5982\u56fe3\u6240\u793a\uff0c\u901a\u8fc7\u5047\u8bbe\u4e86\u4e00\u79cd\u4e24\u9636\u6bb5\u8fc7\u7a0b\u7684\u56e0\u679c\u673a\u5236\uff0c\u6211\u4eec\u5efa\u7acb\u4e86HCR\uff08Hidden Compact Representation\uff09\u6a21\u578b\u3002\u5728\u7b2c\u4e00\u9636\u6bb5\u4e2d\uff0c\u539f\u56e0\u53d8\u91cf\u901a\u8fc7\u4e00\u4e2a\u6052\u7b49\u6620\u5c04\u5f97\u5230\u4e00\u4e2a\u4f4e\u79e9\u7684\u9690\u53d8\u91cf\uff1b\u5728\u7b2c\u4e8c\u9636\u6bb5\u4e2d\uff0c\u7ed3\u679c\u7531\u9690\u53d8\u91cf\u7684\u72b6\u6001\u51b3\u5b9a\uff0c\u5e76\u5728\u968f\u673a\u566a\u58f0\u5e72\u6270\u4e0b\u4ea7\u751f\u3002\u57fa\u4e8e\u4f3c\u7136\u5ea6\u6846\u67b6\uff0c\u5f15 \u5165\u8d1d\u53f6\u65af\u51c6\u5219\uff0c\u7ed9\u51fa\u4e86\u4e00\u79cd\u8bc6\u522b\u8be5\u6a21\u578b\u7684\u65b9\u6cd5\u3002\u56fe 3  HCR \u6a21\u578b 
\u53e6\u4e00\u7c7b\u503c\u5f97\u4eba\u4eec\u5173\u6ce8\u7684\u95ee\u9898\u662f\u73b0\u6709\u65b9\u6cd5\u4e0d\u9002\u7528\u4e8e\u6570\u636e\u542b\u6709\u6d4b\u91cf\u8bef\u5dee\u7684\u60c5\u51b5\u3002\u5728\u73b0\u5b9e\u751f\u6d3b\u4e2d\uff0c\u7531\u4e8e\u89c2\u6d4b\u624b\u6bb5\u7684\u6709\u9650\u6027\uff0c\u6211\u4eec\u6240\u83b7\u53d6\u7684\u6570\u636e\u4e5f\u4e0d\u53ef\u907f\u514d\u5730\u542b\u6709\u6d4b\u91cf\u8bef\u5dee\u3002Scheines\u548cRamsey\u63a2\u7d22\u4e86\u6d4b\u91cf\u8bef\u5dee\u7684\u5b58\u5728\u5bf9\u57fa\u4e8e\u7ebf\u6027\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5\u5f71\u54cd\uff0cZhang\u7b49\u968f\u540e\u63d0\u51fa\u5e76\u8bc1\u660e\u4e86\u542b\u6709\u6d4b\u91cf\u8bef\u5dee\u7684\u7ebf \u6027\u56e0\u679c\u6a21\u578b\u7684\u53ef\u8bc6\u522b\u6761\u4ef6\u3002\u7814\u7a76\u6307\u51fa\uff0c\u5f53\u6570\u636e\u8d8a\u8d8b\u5411\u9ad8\u65af\u5206\u5e03\uff0c\u5bf9\u5e94\u7684\u56e0\u679c\u63a8\u65ad\u4e5f\u4f1a\u8d8a\u56f0\u96be\uff1b\u5f53\u4e0d\u5177\u5907\u5bf9\u6d4b\u91cf\u8bef\u5dee\u7684\u5148\u9a8c\u77e5\u8bc6\u65f6\uff0c\u56e0\u679c\u63a8\u65ad\u4e5f\u662f\u4e0d\u53ef\u80fd\u7684\u3002\u6211\u4eec\u4ece\u5916\u751f\u53d8\u91cf\u7684\u6027\u8d28\u4e2d\u5f97\u5230\u542f\u53d1\uff0c\u63d0\u51fa\u4e00\u79cd\u57fa\u4e8e\u71b5\u7684ETPIA\u7b97\u6cd5\u3002\u7b2c\u4e00\u9636\u6bb5\uff0c\u5229\u7528\u5916\u751f\u53d8\u91cf\u5177\u6709\u71b5\u6700\u5c0f\u7684\u6027\u8d28\u5c06\u8bc6\u522b\u5916\u751f\u53d8\u91cf\u3002\u7b2c\u4e8c\u9636\u6bb5\uff0c\u5254\u9664\u5916\u751f\u53d8\u91cf\u5bf9\u5176\u4f59\u53d8\u91cf\u7684\u5f71\u54cd\u3002\u5728\u5254\u9664\u9636\u6bb5\u9488\u5bf9\u4e0d\u542b\u6709\u6d4b\u91cf\u8bef\u5dee\u7684\u60c5\u51b5\uff0c \u76f4\u63a5\u4f7f\u7528\u56de\u5f52\u7cfb\u6570\u53bb\u9664\u5916 \u751f\u53d8\u91cf\u5bf9\u5176\u4f59\u53d8\u91cf\u7684\u5f71\u54cd\uff1b\u800c\u5728\u542b\u6709\u6d4b\u91cf\u8bef\u5dee\u7684\u60c5\u51b5\u4e0b\uff0c\u5229\u7528\u4f9d\u8d56\u6bd4\u5b9e\u73b0\u5916\u751f\u53d8\u91cf\u6548\u5e94\u7684\u5254\u9664\u3002\u7b2c\u4e09\u9636\u6bb5\uff0c\u5219\u9700\u8981\u5728\u5f97\u5230\u5916\u751f\u53d8\u91cf\u7684\u987a\u5e8f\uff08\u56e0\u679c\u6b21\u5e8f\uff09\u7684\u57fa\u7840\u4e0a\u4f7f\u7528\u526a\u679d\u7b97\u6cd5\u83b7\u5f97\u6700\u7ec8\u7684\u56e0\u679c\u7f51\u7edc\u3002
\u4e0a\u8ff0\u5185\u5bb9\u90fd\u662f\u8003\u8651\u4ece\u89c2\u5bdf\u6570\u636e\u4e2d\u53d1\u73b0\u56e0\u679c\u5173\u7cfb\uff0c\u5e76\u4e0d\u9002\u7528\u4e8e\u542b\u6709\u9690\u53d8\u91cf\u7684\u60c5\u51b5\u3002\u7279\u522b\u5730\uff0c\u5728\u9690\u53d8\u91cf\u7684\u7814\u7a76\u4e2d\uff0c\u5982\u4f55\u5b66\u4e60\u9690\u53d8\u91cf\u4e4b\u95f4\u7684\u56e0\u679c\u5173\u7cfb\uff0c\u662f\u5f53\u4eca\u7814\u7a76\u7684\u70ed\u70b9\u95ee\u9898\u548c\u6311\u6218\u3002\u5728LiNGAM\u6a21\u578b\u7684\u57fa\u7840\u4e0a\uff0c\u901a\u8fc7\u5f15\u5165\u9690\u53d8\u91cf\uff0cTashiro\u548c Shimizu\u7b49\u63d0\u51fa\u4e86ParceLiNGAM\u7b97\u6cd5\uff0c\u4e3b\u8981\u901a\u8fc7\u68c0\u9a8c\u4f30\u8ba1\u56de\u5f52\u6b8b\u5dee\u4e0e\u5916\u751f\u53d8\u91cf\u7684\u72ec\u7acb\u6027\u548c\u627e\u5230\u5305\u542b\u672a\u88ab\u9690\u53d8\u91cf\u6240\u5f71\u54cd\u7684\u53d8\u91cf\u5b50\u96c6\u6765\u53d1\u73b0\u9690\u53d8 \u91cf\uff1bHoyer\u7b49\u7ed3\u5408LiNGAM\u6a21\u578b\uff0c\u63d0\u51fa\u9002\u7528\u4e8e \u7ebf\u6027\u975e\u9ad8\u65af\u6761\u4ef6\u4e0b\u7684lvLiNGAM\uff08latent variable LiNGAM\uff09\u6846\u67b6\u3002\u4f46\u662f\u8fd9\u4e9b\u7814\u7a76\u5927\u591a\u6570\u5173\u6ce8\u4e8e\u5728\u542b\u6709\u9690\u53d8\u91cf\u7684\u60c5\u51b5\u4e0b\u53d1\u73b0\u53ef\u89c2\u5bdf\u53d8\u91cf\u7684\u56e0\u679c\u7ed3\u6784\uff0c\u800c\u4e0d\u5728\u4e8e\u53d1\u73b0\u9690\u53d8\u91cf\u7684\u56e0\u679c\u7ed3\u6784\u3002\u5982\u65e8\u5728\u53d1\u73b0\u9690\u53d8\u91cf\u7ed3\u6784\u7684\u5de5\u4f5c\uff08\u5982Tetrad\uff09\uff0c\u5f80\u5f80\u9700\u8981\u66f4\u591a\u7684\u53ef\u89c2\u5bdf\u53d8\u91cf\uff0c\u4e14\u5b83\u4eec\u8f93\u51fa\u7684\u662f\u4e00\u4e2a\u7b49\u4ef7 \u7c7b\u3002\u4e3a\u4e86\u5728\u89c2\u5bdf\u53d8\u91cf\u6570\u91cf\u4e0d\u591a\u7684\u60c5\u51b5\u4e0b\u8bc6\u522b\u9690\u53d8\u91cf\u7ed3\u6784\uff0c\u901a\u8fc7\u5f15\u5165\u975e\u9ad8\u65af\u6027\u5047\u8bbe\uff0c\u6211\u4eec\u63d0\u51fa\u4e86\u4e00\u79cd\u57fa\u4e8eTriad\u7ea6\u675f\u6761\u4ef6\u7684\u9690\u53d8\u91cf\u56e0\u679c\u7ed3\u6784\u5b66\u4e60\u7b97\u6cd5 LSTC\uff08Learn the Structure of latent variables based on Triad Constraints\uff09[28]\u3002\u5728\u5176\u4ed6\u7c7b\u578b\u6570\u636e\u4e0a\uff0cZhang\u7b49\u6269\u5c55\u4e86GPLVM\uff08Gaussian-Process Latent Variable Model\uff09\uff0c\u63d0\u51fa\u4e86IGPLVM\uff08Invariant Gaussian Process Latent Variable Models\uff09\u7b97\u6cd5\u6765\u5904\u7406\u9690\u53d8\u91cf\u5bf9\u89c2\u5bdf\u53d8\u91cf\u7684\u56e0\u679c\u4f5c\u7528\u662f\u975e\u7ebf\u6027\u7684\uff0c\u800c\u89c2\u5bdf\u53d8\u91cf\u95f4\u7684\u56e0\u679c\u4f5c\u7528\u662f\u7ebf\u6027\u7684\u60c5\u51b5\u3002
1.3 \u6df7\u5408\u578b\u65b9\u6cd5
\u6df7\u5408\u578b\u65b9\u6cd5\u662f\u878d\u5408\u4e86\u57fa\u4e8e\u7ea6\u675f\u7684\u65b9\u6cd5\u548c\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5\u800c\u53d1\u5c55\u51fa\u6765\u7684\u53e6\u4e00\u7c7b\u65b9\u6cd5\u3002\u8fd9\u7c7b\u65b9\u6cd5\u6709\u6548\u5730\u63d0\u9ad8\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u4e0d\u8db3\uff0c\u540c\u65f6\u514b\u670d\u4e86\u9ad8\u7ef4\u6570\u636e\u4e0a\u8bef\u53d1\u73b0\u7387\u63a7\u5236\u96be\u9898\u3002\u73b0\u6709\u7684\u6df7\u5408\u578b\u65b9\u6cd5\u4e3b\u8981\u4f9d\u8d56\u5206\u6cbb\u7b56\u7565\u3001\u7ec4\u88c5\u7b56\u7565\u4e0e\u878d\u5408\u7b56\u7565\u8fd9\u4e09\u7c7b\u7b56\u7565\u65b9\u6cd5\u3002
\u5229\u7528\u5206\u6cbb\u7b56\u7565\uff0c\u6211\u4eec\u5c1d\u8bd5\u5c06\u56e0\u679c\u63a8\u65ad\u95ee\u9898\u5206\u89e3\u4e3a\u5b50\u95ee\u9898\u5e76\u5229\u7528\u9012\u5f52\u65b9\u6cd5\u6c42\u89e3\uff0c\u63d0\u51fa\u4e86SADA\uff08Scalable cAusation Discovery Algorithm\uff09\u6846\u67b6\uff0c\u4e3b\u8981\u9002\u7528\u4e8e\u56e0\u679c\u7ed3\u6784\u4e2d\u7684\u7a00\u758f\u5c5e\u6027\u7684\u89c2\u5bdf\u5206\u6790\uff0c\u5728\u6837\u672c\u96c6\u8f83\u5c11\u7684\u60c5\u51b5\u4e0b\u4e5f\u80fd\u6b63\u786e\u5730\u8bc6\u522b\u56e0\u679c\u53d8\u91cf\u3002\u5176\u4e3b\u8981\u601d\u60f3\u662f\uff0c\u9996\u5148\u901a\u8fc7\u6c42\u89e3\u56e0\u679c\u5206\u5272\u96c6\u5c06\u9ad8\u7ef4\u95ee\u9898\u5206\u89e3\u62102\u4e2a\u5b50\u95ee\u9898\uff1b\u7136\u540e\u9488\u5bf9\u6bcf\u4e2a\u5b50\u95ee\u9898\u8fdb\u884c\u9012\u5f52\u5206\u89e3\u76f4\u5230\u5176\u95ee\u9898\u89c4\u6a21\u8db3\u591f\u5c0f\uff1b\u9488\u5bf9\u6bcf\u4e2a\u8db3\u591f\u5c0f\u7684\u5b50\u95ee\u9898\uff0c\u91c7\u7528ANM\u7b49\u57fa\u4e8e\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5\u8fdb\u884c\u6c42\u89e3\uff0c\u6700\u540e\u5bf9\u5c0f\u95ee\u9898\u8fdb\u884c\u5408\u5e76\u3002
\u6211\u4eec\u5728\u7814\u7a76\u4e2d\u53d1\u73b0\uff0c\u5206\u6cbb\u7b56\u7565\u5728\u5206\u89e3\u95ee\u9898\u65f6\u5f15\u5165\u9519\u8bef\u7684\u5212\u5206\uff0c\u5728\u540e\u7eed\u8fc7\u7a0b\u4e2d\u4f1a\u4e0d\u65ad\u7d2f\u79ef\u4f7f\u5f97\u603b\u4f53\u8bef\u5dee\u5448\u73b0\u67d0\u79cd\u4e0d\u53ef\u63a7\u7684\u73b0\u8c61\u3002\u800c\u7ec4\u88c5\u7b56\u7565\u53ef\u4ee5\u9488\u5bf9\u968f\u673a\u5c0f\u53d8\u91cf\u96c6\u5408\uff0c\u901a\u8fc7\u67d0\u79cd\u590d\u6742\u7684\u805a\u5408\u8fc7\u7a0b\u6392\u9664\u7531\u4e8e\u5206\u5272\u5f15\u5165\u7684\u7ed3\u6784\u9519\u8bef\uff0c\u83b7\u5f97\u53ef\u9760\u56e0\u679c\u673a\u5236\u3002\u6240\u4ee5\u6211\u4eec\u8bbe\u8ba1\u4e86SMRP\uff08Sophisticated Merging over Random Partitions\uff09\u7b97\u6cd5\u6765\u5408\u5e76\u6240\u6709\u5212\u5206\u7684\u7ed3\u679c\uff0c\u8fd0\u7528\u57fa\u4e8e\u4f20\u64ad\u7684\u663e\u8457\u6027\u589e\u5f3a\u65b9\u6cd5\u548c\u6700\u5927\u65e0\u73af\u5b50\u56fe\u7684\u56e0\u679c\u6b21\u5e8f\u65b9\u6cd5\u7b49\u5bf9\u5c40\u90e8\u7ed3\u679c\u8fdb\u884c\u5408\u5e76\u3002\u8be5\u6846\u67b6\u80fd\u901a\u8fc7\u53ef\u9760\u7684\u56e0\u679c\u673a\u5236\uff0c\u6709\u6548\u5730\u5408\u5e76\u968f\u673a\u5206\u5757\u7684\u90e8\u5206\u7ed3\u679c\u3002
\u4e0a\u8ff0\u4e24\u7c7b\u7b56\u7565\u53ca\u65b9\u6cd5\u4e3b\u8981\u57fa\u4e8e\u5206\u6cbb\u7684\u601d\u60f3\uff0c\u8fd8\u6709\u53e6\u4e00\u7c7b\u8003\u8651\u878d\u5408\u4e0d\u540c\u65b9\u6cd5\u7684\u7b56\u7565\u3002\u8003\u8651\u5230\u57fa\u4e8e\u8bc4\u5206\u7684\u65b9\u6cd5\u5f97\u5230\u7684\u7ed3\u679c\u5b58\u5728\u9a6c\u5c14\u53ef\u592b\u7b49\u4ef7\u7c7b\u95ee\u9898\uff0c\u800c\u57fa\u4e8e\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5\u6709\u52a9\u4e8e\u6d88\u9664\u9a6c\u5c14\u53ef\u592b\u7b49\u4ef7\u7c7b\uff0c\u6545\u5c1d\u8bd5\u5c06\u4e24\u8005\u8fdb\u884c\u878d\u5408\uff0c\u63d0\u51fa\u4e86SELF\uff08Structural Equational Likelihood Framework\uff09\u6846\u67b6\u3002\u5176\u6838\u5fc3\u601d\u60f3\u662f\u5c06\u56e0\u679c\u51fd\u6570\u7684\u566a\u58f0\u72ec\u7acb\u6027\u5047\u8bbe\u5d4c\u5165\u4f3c\u7136\u5ea6\u8ba1\u7b97\u4e2d\uff0c\u901a\u8fc7\u4f3c\u7136\u5ea6\u6846\u67b6\u5b9e\u73b0\u4e24\u7c7b\u65b9\u6cd5\u7684\u7edf\u4e00\u3002\u5728\u542b\u6709\u591a\u4e2a\u9690\u6df7\u6dc6\u56e0\u5b50\u7684\u60c5\u51b5\u4e0b\uff0c\u6211\u4eec\u63d0\u51fa\u4e86\u878d\u5408\u57fa\u4e8e\u7ea6\u675f\u65b9\u6cd5\u548c\u57fa\u4e8e\u56e0\u679c\u51fd\u6570\u6a21\u578b\u65b9\u6cd5\u7684MLCLiNGAM\uff08LiNGAM with Multiple Latent Confounders\uff09 \u7b97\u6cd5\u3002\u8be5\u65b9\u6cd5\u80fd\u591f\u5feb\u901f\u68c0\u6d4b\u5230\u53d7\u9690\u6df7\u6dc6\u56e0\u5b50\u5f71\u54cd\u7684\u89c2\u5bdf\u53d8\u91cf\uff0c\u6709\u52a9\u4e8e\u89e3\u51b3\u542b\u6709\u591a\u5143\u9690\u6df7\u6dc6\u56e0\u5b50\u7684\u56e0\u679c\u7ed3\u6784\u5b66\u4e60\u56f0\u96be\u7684\u95ee\u9898\u3002
\u6df7\u5408\u578b\u65b9\u6cd5\u4e00\u5b9a\u7a0b\u5ea6\u5b9e\u73b0\u4e86\u57fa\u4e8e\u7ea6\u675f\u7684\u65b9\u6cd5\u7684\u9ad8\u7ef4\u6269\u5c55\u6027\u548c\u57fa\u4e8e\u56e0\u679c\u51fd\u6570\u6a21\u578b\u7684\u65b9\u6cd5\u7684\u56e0\u679c\u53d1\u73b0\u80fd\u529b\u7684\u7ed3\u5408\uff0c\u4e3a\u6211\u4eec\u5728\u9ad8\u7ef4\u6570\u636e\u573a\u666f\u4e2d\u7684\u5e94\u7528\u63d0\u4f9b\u4e86\u91cd\u8981\u57fa\u7840\u3002\u4f8b\u5982\u5728\u4e0e\u4e2d\u56fd\u5357\u65b9\u7535\u7f51\u5408\u4f5c\u4e2d\uff0c\u901a\u8fc7\u8fd0\u7528\u6df7\u5408\u578b\u65b9\u6cd5\u5efa\u7acb\u4e86\u7535\u7f51\u4fe1\u606f\u5b50\u7cfb\u7edf\u7684\u6545\u969c\u53d1\u751f\u6a21\u578b\uff0c\u5e76\u57fa\u4e8e\u6545\u969c\u56e0\u679c\u6eaf\u6e90\uff0c\u5b9e\u73b0\u4e86\u6839\u56e0\u6545\u969c\u7684\u5feb\u901f\u5b9a\u4f4d\u3002\u5728\u8be5\u5e94\u7528\u6848\u4f8b\u4e2d\uff0c\u6545\u969c\u5b9a\u4f4d\u4e2d\u7684\u5e73\u5747\u51c6\u786e\u7387\u7531\u539f\u6765\u768455.56%\u63d0\u9ad8\u5230 91.67%\uff0c\u5927\u5927\u51cf\u5c11\u4e86\u6545\u969c\u6392\u67e5\u8303\u56f4\uff0c\u63d0\u5347\u4e86\u7cfb\u7edf\u8fd0\u884c\u7684\u53ef\u9760\u6027\u3002\u6211\u4eec\u8fd8\u4e0e\u5357\u65b9\u901a\u8baf\u5efa\u8bbe\u3001\u534e\u4e3a\u7b49\u5355\u4f4d\u5408\u4f5c\uff0c\u5728\u79fb\u52a8\u901a\u8baf\u7f51\u7edc\u57fa\u7ad9\u6027\u80fd\u4f18\u5316\u4e0a\u8fdb\u884c\u4e86\u5e94\u7528\uff0c\u901a\u8fc7\u91cd\u6784\u57fa\u7ad9\u6027\u80fd\u6307\u6807\u4e4b\u95f4\u7684\u56e0\u679c\u7f51\u7edc\uff0c\u6210\u529f\u7ed9\u51fa\u4e86\u6570\u636e\u8fde\u63a5\u6027\u80fd\u548c\u901a\u8bdd\u8d28\u91cf\u4f18\u5316\u7b49\u91cd\u8981\u5178\u578b\u6295\u8bc9\u7684\u4f18\u5316\u65b9\u6848\uff0c\u76f8\u5173\u65b9\u6848\u5728\u5b9e\u9645\u5e94\u7528\u6548\u679c\u4e2d\u8d85\u8fc7\u4e86\u4f20\u7edf\u9886\u57df\u4e13\u5bb6\u57fa\u4e8e\u7ecf\u9a8c\u7ed9\u51fa\u7684\u4f18\u5316\u65b9\u6848\u3002\u8fd9\u4e9b\u6210\u529f\u5e94\u7528\u6848\u4f8b\u4f53\u73b0\u4e86\u56e0\u679c\u63a8\u65ad\u5728\u51b3\u7b56\u652f\u6301\u9886\u57df\u7684\u91cd\u8981\u4ef7\u503c\uff0c\u662f\u56e0\u679c\u63a8\u65ad\u7814\u7a76\u7684\u91cd\u8981\u65b9\u5411\u3002
 <\/p>\n

2 \u56e0\u679c\u6027\u5b66\u4e60<\/p>\n

\u76ee\u524d\u4ee5\u6df1\u5ea6\u5b66\u4e60\u4e3a\u4ee3\u8868\u7684\u673a\u5668\u5b66\u4e60\u6b63\u53d7\u5230\u5b66\u8005\u4eec\u8d8a\u6765\u8d8a\u591a\u7684\u5173\u6ce8\u3002\u7136\u800c\uff0c\u673a\u5668\u5b66\u4e60\uff0c\u5c24\u5176\u662f\u6df1\u5ea6\u5b66\u4e60\u7684\u53ef\u89e3\u91ca\u6027\u3001\u6cdb\u5316\u80fd\u529b\u548c\u5bf9\u6570\u636e\u7684\u8fc7\u5ea6\u4f9d\u8d56\u662f\u76ee\u524d\u516c\u8ba4\u7684\u6311\u6218\u3002\u4e3a\u6b64\uff0c\u5b66\u754c\u8d8a\u6765\u8d8a\u5173\u6ce8\u5728\u673a\u5668\u5b66\u4e60\u4e2d\u56e0\u679c\u601d\u60f3\u7684\u8fd0\u7528\uff0c\u5728\u534a\u76d1\u7763\u5b66\u4e60\uff08SSL\uff09\u548c\u9886\u57df\u81ea\u9002\u5e94\u7b49\u65b9\u9762\u8fdb\u884c\u4e86\u5c1d\u8bd5\u3002\u76f8\u5173\u7814\u7a76\u8868\u660e\uff0c\u56e0\u679c\u63a8\u65ad\u7406\u8bba\u7ed9\u51fa\u4e86\u9690\u85cf\u5728\u89c2\u5bdf\u6570\u636e\u80cc\u540e\u7684\u6709\u7528\u4fe1\u606f\uff0c\u4e3a\u534a\u76d1\u7763\u5b66\u4e60\u548c\u9886\u57df\u81ea\u9002\u5e94\u7b49\u673a\u5668\u5b66\u4e60\u9886\u57df\u7684\u7814\u7a76\u63d0\u4f9b\u4e86\u65b0\u601d\u8def\u548c\u65b9\u5411\u3002\u6211\u4eec\u5bf9\u8fd9\u4e24\u65b9\u9762\u7684\u56e0\u679c\u6027\u5b66\u4e60\u65b9\u6cd5\u8fdb\u884c\u4e86\u603b\u7ed3\uff0c\u5982\u88682\u6240\u793a\u3002<\/p>\n

\u88682  \u56e0\u679c\u6027\u5b66\u4e60\u76f8\u5173\u65b9\u6cd5

\u5728\u534a\u76d1\u7763\u5b66\u4e60\u4e2d\uff0c\u9a6c\u666e\u6240\u7684Sch\u00f6lkopf\u7b49\u6307\u51fa\u534a\u76d1\u7763\u5b66\u4e60\u5728\u56e0\u679c\u65b9\u5411\u4e0a\u7684\u5b66\u4e60\u4e0e\u53cd\u56e0\u679c\u65b9\u5411\u4e0a\u5b66\u4e60\u7684\u533a\u522b\uff0c\u63ed\u793a\u4e86\u5728\u6ca1\u6709\u6df7\u6dc6\u56e0\u5b50\u7684\u60c5\u51b5\u4e0b\uff0c\u65e0\u6807\u7b7e\u6570\u636e\u53ea\u6709\u5728\u53cd\u56e0\u679c\u65b9\u5411\u4e0b\u624d\u662f\u6709\u6548\u7684\uff0c\u800c\u5728\u56e0\u679c\u65b9\u5411\u4e0a\u662f\u65e0\u6548\u7684\u3002\u4ed6\u4eec\u53d1\u73b0\uff0c\u534a\u76d1\u7763\u5b66\u4e60\u65b9\u6cd5\u7684\u6709\u6548\u6027\u4e0e\u56e0\u679c\u5173\u7cfb\u4e2d\u539f\u56e0\u53d8\u91cf\u7684\u6982\u7387P\uff08\u539f\u56e0\uff09\uff0c\u4e0e\u7ed9\u5b9a\u539f\u56e0\u53d8\u91cf\u7684\u60c5\u51b5\u4e0b\u7ed3\u679c\u53d8\u91cf\u7684\u6982\u7387 P\uff08\u7ed3\u679c|\u539f\u56e0\uff09\u7684\u72ec\u7acb\u6027\u6709\u7d27\u5bc6\u8054\u7cfb\u3002 
\u5728\u9886\u57df\u81ea\u9002\u5e94\u7814\u7a76\u4e2d\uff0c\u786e\u5b9a\u4ece\u6e90\u57df\u8f6c\u79fb\u5230\u76ee\u6807\u9886\u57df\u7684\u4fe1\u606f\uff0c\u4ee5\u53ca\u5982\u4f55\u8fdb\u884c\u8f6c\u79fb\u662f\u4e00\u4e2a\u81f3\u5173\u91cd\u8981\u7684\u95ee\u9898\u3002\u56e0\u679c\u6a21\u578b\u53ef\u4ee5\u4f5c\u4e3a\u4e00\u4e2a\u5f88\u597d\u7684\u5de5\u5177\u6765\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\u3002Pearl\u6559\u6388\u901a\u8fc7\u7814\u7a76\u8bc1\u660e\u56e0\u679c\u56fe\u53ef\u4ee5\u88ab\u7528\u4e8e\u5efa\u7acb\u5141\u8bb8\u8de8\u57df\u7684\u7ed3\u679c\u4f20\u8f93\u7684\u6761\u4ef6\uff1bZhang\u7b49\u53d1\u73b0\u5373\u4f7f\u5728\u8fd9\u79cd\u6761\u4ef6\u4e0d\u6210\u7acb\u7684\u60c5\u51b5\u4e0b\uff0c\u4ecd\u7136\u53ef\u4ee5\u5229\u7528\u56e0\u679c\u77e5\u8bc6\uff0c\u4ee5\u53ca\u4e00\u4e9b\u6280\u672f\u6761\u4ef6\u6765\u8fdb\u884c\u9886\u57df\u81ea\u9002\u5e94\u3002\u5176\u57fa\u672c\u601d\u60f3\u662f\uff0c\u5982\u679c \u5b83\u4eec\u4e4b\u95f4\u6ca1\u6709\u6df7\u6742\u56e0\u5b50\uff0c\u5219P\uff08\u539f\u56e0\uff09\u548cP\uff08\u7ed3\u679c|\u539f\u56e0\uff09\u662f\u771f\u5b9e\u56e0\u679c\u8fc7\u7a0b\u7684\u53cd\u6620\uff0c\u5e76\u4e14\u53d8\u5316\u662f \u72ec\u7acb\u7684\uff0c\u4ece\u800c\u5141\u8bb8\u4ee5\u7b80\u5355\u7684\u5f62\u5f0f\u5206\u522b\u5bf9\u53d8\u5316\u8fdb\u884c \u53c2\u6570\u5316\u3002\u5f53\u6e90\u57df\u662f\u591a\u4e2a\u7684\u60c5\u51b5\u4e0b\uff0cZhang\u7b49\u5c06\u5df2 \u77e5\u7684\u56e0\u679c\u5173\u7cfb\u77e5\u8bc6\uff08\u6570\u636e\u80cc\u540e\u4ea7\u751f\u7684\u673a\u5236\uff09\u878d\u5408\u5230\u9886\u57df\u81ea\u9002\u5e94\u4e2d\uff0c\u63d0\u51fa\u4e86\u4e09\u79cd\u9002\u7528\u4e8e\u4e0d\u540c\u573a\u666f\u7684\u65b9\u6cd5\uff0c\u5373\u5229\u7528\u91cd\u8981\u6027\u6743\u91cd\u8c03\u6574\u7684weigh_sample\u7b97\u6cd5\u3001\u4f7f\u7528\u4e86\u901a\u7528\u6027\u5b66\u4e60\u7684genar_model\uff0c\u4ee5\u53ca\u501f\u52a9\u4e86\u6e90\u5047\u8bbe\u7684\u6743\u91cd\u7ed3\u5408\u7684combn_classf\u65b9\u6cd5\u3002\u6211\u4eec\u5219\u5206\u522b\u4ece\u56e0\u679c\u89e3\u8026\u548c\u56e0\u679c\u540c\u6001\u5047\u8bbe\u8fd9\u4e24\u4e2a\u89d2\u5ea6\u63a2\u8ba8\u4e86\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u3002\u4ece\u56e0\u679c\u89e3\u8026\u89d2\u5ea6\uff0c\u5047\u8bbe\u4e0d\u540c\u9886\u57df\u7684\u6570\u636e\u7531\u9886\u57df\u9690\u53d8\u91cf\u548c\u8bed\u4e49\u9690\u53d8\u91cf\u7ec4\u6210\uff0c\u5176\u56e0\u679c\u56fe\u8868\u793a\u89c1\u56fe4\u3002\u57fa\u4e8e\u8fd9\u4e2a\u5047\u8bbe\uff0c\u6211\u4eec\u63d0\u51fa\u4e86\u8bed\u4e49\u89e3\u8026\u8868\u8fbe\uff08DSR\uff09\u6a21\u578b\uff0c\u91c7\u7528\u53d8\u5206\u81ea\u52a8\u7f16\u7801\u673a\u548c\u5bf9\u68af\u5ea6\u53cd\u8f6c\u5b66\u4e60\u65b9\u6cd5\u5b9e\u73b0\u4e86\u9886\u57df\u9690\u53d8\u91cf\uff08Zd\uff09\u548c\u8bed\u4e49\u9690\u53d8\u91cf\uff08Zy\uff09\u7684\u91cd\u6784\u548c\u89e3\u8026\u3002\u5728\u56e0\u679c\u540c\u6001\u5047\u8bbe\u4e2d\uff0c\u5047\u8bbe\u4e0d\u540c\u9886\u57df\u6570\u636e\u670d\u4ece\u540c\u6001\u5047\u8bbe\uff0c\u5373\u5171\u4eab\u90e8\u5206\u76f8\u540c\u56e0\u679c\u7f51\u7edc\u7ed3\u6784\uff08\u89c1\u56fe5\uff09\u3002\u901a\u8fc7\u5b66\u4e60\u4e0d\u540c\u9886\u57df\u6570\u636e\u95f4\u7684\u540c\u6001\u6620\u5c04\u548c\u540c\u6001\u7b97\u5b50\uff0c\u6211\u4eec\u63d0\u51faDACH\u65b9\u6cd5\u6210\u529f\u5206 \u79bb\u51fa\u4e86\u4e0d\u540c\u9886\u57df\u80cc\u540e\u5171\u4eab\u7684\u56e0\u679c\u673a\u5236\uff0c\u5b9e\u73b0\u4e86\u53ef\u9760\u7684\u9886\u57df\u81ea\u9002\u5e94\u3002\u6b64\u5916\uff0c\u8fd8\u57fa\u4e8e\u7279\u5b9a\u7684\u5047\u8bbe\uff0c\u4ece\u7406\u8bba\u89d2\u5ea6\u7ed9\u51fa\u4e86\u9886\u57df\u81ea\u9002\u5e94\u95ee\u9898\u8bef\u5dee\u754c\u3002\u56fe 4  \u4e0d\u540c\u9886\u57df\u6570\u636e\u751f\u6210\u8fc7\u7a0b\u56e0\u679c\u56fe
\u56fe 5  \u56e0\u679c\u540c\u6001\u793a\u610f\u56fe\uff08\u5728\u5343\u514b\u548c\u78c5\u4e24\u4e2a\u9886\u57df\u4e2d\u8fd0\u52a8\u524d \u4f53\u91cd\u5927\u4e8e\u8fd0\u52a8\u540e\u4f53\u91cd\u8fd9\u4e00\u8fd0\u7b97\u7ed3\u679c\u4fdd\u6301\u4e0d\u53d8\uff09 <\/p>\n

 <\/p>\n

3 \u7ed3\u675f\u8bed<\/p>\n

\u672c\u6587\u5bf9\u56e0\u679c\u63a8\u65ad\u53ca\u56e0\u679c\u6027\u5b66\u4e60\u7684\u4e00\u4e9b\u57fa\u672c\u65b9\u6cd5\u548c\u6700\u65b0\u8fdb\u5c55\u8fdb\u884c\u7b80\u8981\u4ecb\u7ecd\u3002\u76ee\u524d\u56e0\u679c\u63a8\u65ad\u9886\u57df\u7814\u7a76\u5df2\u7ecf\u6d8c\u73b0\u51fa\u5927\u91cf\u76f8\u5173\u65b9\u6cd5\uff0c\u5e76\u5f00\u59cb\u5728\u6839\u56e0\u6545\u969c\u5b9a\u4f4d\u7b49\u9886\u57df\u663e\u793a\u51fa\u826f\u597d\u7684\u5e94\u7528\u6548\u679c\u3002\u4f46\u662f\uff0c\u56e0\u679c\u7b49\u4ef7\u7c7b\u7684\u65b9\u5411\u5224\u522b\u3001\u9ad8\u7ef4\u6570\u636e\u4e0a\u7684\u8bef\u53d1\u73b0\u7387\u63a7\u5236\u548c\u4e0d\u5b8c\u5168\u89c2\u5bdf\u6570\u636e\u4e0a\u7684\u9690\u53d8\u91cf\u68c0\u6d4b\u7b49\u96be\u9898\u4ecd\u6709\u5f85\u8fdb\u4e00\u6b65\u89e3\u51b3\u3002\u56e0\u679c\u6027\u5b66\u4e60\u65b9\u9762\u7814\u7a76\u5219\u8fd8\u5904\u4e8e\u8d77\u6b65\u9636\u6bb5\uff0c\u672a\u6765\u8fd8\u6709\u5f88\u5927\u7684\u53d1\u5c55\u7a7a\u95f4\u3002\u5f53\u7136\uff0c\u56e0\u679c\u6027\u5b66\u4e60\u7684\u53d1\u5c55\u4e00\u5b9a\u7a0b\u5ea6\u4e5f\u53d7\u9650\u4e8e\u56e0\u679c\u63a8\u65ad\u7406\u8bba\u4e0e \u65b9\u6cd5\u7684\u7a81\u7834\u3002\u4f8b\u5982\uff0c\u73b0\u6709\u673a\u5668\u5b66\u4e60\u4efb\u52a1\u4e2d\u5f88\u96be\u4fdd \u8bc1\u6570\u636e\u7684\u5b8c\u5168\u89c2\u5bdf\u7279\u6027\uff0c\u9650\u5236\u4e86\u56e0\u679c\u63a8\u65ad\u7406\u8bba\u4e0e \u65b9\u6cd5\u7684\u5e94\u7528\u3002\u7efc\u4e0a\uff0c\u56e0\u679c\u63a8\u65ad\u548c\u56e0\u679c\u6027\u5b66\u4e60\u662f\u503c\u5f97\u6df1\u5165\u7814\u7a76\u7684\u4efb\u52a1\u3002 
\u4f5c\u8005   \u8521\u745e\u521d\uff0c\u5e7f\u4e1c\u5de5\u4e1a\u5927\u5b66\u6559\u6388\u3002\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u56e0\u679c\u63a8\u65ad\u3001\u56fe\u795e\u7ecf\u7f51\u7edc\u548c\u8fc1\u79fb\u5b66\u4e60\u7b49\u3002\u9648\u8587\uff0c\u5e7f\u4e1c\u5de5\u4e1a\u5927\u5b66\u535a\u58eb\u7814\u7a76\u751f\u3002\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u4e3a\u56e0\u679c\u63a8\u65ad\u3002\u90dd\u5fd7\u5cf0\uff0c\u4f5b\u5c71\u79d1\u5b66\u6280\u672f\u5b66\u9662\u6821\u957f\u3001\u6559\u6388\u3002\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u4e3a\u4f18\u5316\u7b97\u6cd5\u548c\u673a\u5668\u5b66\u4e60\u7b49\u3002<\/p>\n


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

[1] Zhang K, Sch\u00f6lkopf B, Spirtes P, et al. Learning causality and causality-related learning: some recent progress[J]. National science review, 2018, 5(1): 26-29.[2]  Spirtes P, Glymour C N, Scheines R. Causation, prediction, and search, second edition. Cambridge, USA: MIT press, 2000.[3] Verma T, Pearl J. Equivalence and synthesis of causal models\/\/ Proceedings of the 6th Conference on Uncertainty in Artificial Intelligence. Cambridge, UK, 1990: 255-268.[4]  He Y, Geng Z. Active learning of causal networks with intervention experiments and optimal designs. Journal of Machine Learning Research, 2008, 9(11):2523-2547.[5] Yin J, Zhou Y, Wang C, et al. Partial orientation and local structural learning of causal networks for prediction\/\/Proceedings of the WCCI Causation and Prediction Challenge. Hong Kong, China, 2008: 93-105.[6] Zhou Y, Wang C, Yin J, Geng Z. Discover local causal network around a target to a given depth\/\/Guyon Isabelle, Janzing Dominik, Sch\u00f6lkopf Bernhard eds. NIPS Causality: Objectives and Assessment. Cambridge, USA: MIT Press, 2010: 191-202.[7] Tsamardinos I, Brown L E, Aliferis C F. The max-min hill-climbing Bayesian network structure learning algorithm. Machine Learning, 2006, 65(1): 31-78.[8] Tsamardinos I, Aliferis C F, Statnikov A. Time and sample efficient discovery of Markov blankets and direct causal relations\/\/Proceedings of the Ninth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Washington, USA, 2003: 673-678.[9] Spirtes P, Meek C, Richardson T. An algorithm for causal inference in the presence of latent variables and selection bias.\/\/Cooper G F, Glymour C eds. IEEE Transactions on Pattern Analysis and Machine Intelligence. Cambridge, USA: MIT Press, 1999: 211\u2013252.[10] Colombo D, Maathuis M H, Kalisch M, et al. Learning high-dimensional directed acyclic graphs with latent and selection variables. The Annals of Statistics, 2012, 40(1): 294-321.[11] Ogarrio J M, Spirtes P, Ramsey J. A hybrid causal search algorithm for latent variable models[C]\/\/Conference on Probabilistic Graphical Models. 2016: 368-379.[12] Kummerfeld E, Ramsey J, Yang R, et al. Causal clustering for 2-factor measurement models\/\/Proceedings of the Joint European Conference on Machine Learning and Knowledge Discovery in Databases (ECML PKDD 2014). Nancy, France, 2014: 34-49.[13] Shimizu S, Hoyer P O, Hyv\u00e4rinen A, et al. A linear non-Gaussian acyclic model for causal discovery. Journal of Machine Learning Research, 2006, 7: 2003-2030.[14] Zhang K, Hyv\u00e4rinen A. On the identifiability of the post-nonlinear causal model\/\/Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence (UAI 2009). Montreal, Canada, 2009: 647-655.[15] Zhang K, Chan L. Extensions of ICA for Causality Discovery in the Hong Kong Stock Market\/\/Proceedings of the 13th International Conference on Neural Information (ICONIP). Hong Kong, China, 2006: 400-409.[16] Hoyer P O, Janzing D, Mooij J M, et al. Nonlinear causal discovery with additive noise models\/\/Proceedings of the 23rd Annual Conference on Neural Information Processing Systems (NIPS 2009). Vancouver, Canada, 2009: 689-696.[17] Peters J, Janzing D, Sch\u00f6lkopf B. Causal inference on discrete data using additive noise models. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(12): 2436-2450.[18] Daniusis, P., Janzing, D., Mooij, J., Zscheischler, J., Steudel, B., Zhang, K., Sch\u00f6lkopf, B. Inferring deterministic causal relations\/\/Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence (UAI 2010). Catalina Island, USA, 2010: 143-150.[19] Janzing D, Mooij J, Zhang Kun, et al. Information-geometric approach to inferring causal directions. Artificial Intelligence, 2012, 182: 1-31.[20] Cai R, Qiao J, Zhang K, et al. Causal Discovery with Cascade Nonlinear Additive Noise Model[C]\/\/Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence. 2019.[21] Cai R, Qiao J, Zhang K, et al. Causal discovery from discrete data using hidden compact representation[C]\/\/Advances in neural information processing systems. 2018: 2666-2674.[22] Shimizu S, Inazumi T, Sogawa Y, et al. DirectLiNGAM: A direct method for learning a linear non-gaussian structural equation model. Journal of Machine Learning Research, 2011, 12(2): 1225-1248.[23] Xie F, Cai R, Zeng Y, et al. An Efficient Entropy-Based Causal Discovery Method for Linear Structural Equation Models With IID Noise Variables[J]. IEEE transactions on neural networks and learning systems, 2019.[24] Tashiro T, Shimizu S, Hyv\u00e4rinen A, et al. ParceLiNGAM: A causal ordering method robust against latent confounders. Neural Computation, 2014, 26(1): 57-83.[25] Hoyer P O, Shimizu S, Kerminen A J, et al. Estimation of causal effects using linear non-Gaussian causal models with hidden variables. International Journal of Approximate Reasoning, 2008, 49(2): 362-378.[26] Zhang K, Sch\u00f6lkopf B., Janzing D. Invariant Gaussian process latent variable models and application in causal discovery\/\/Proceedings of the Twenty-Sixth Conference (UAI 2010). Corvallis, USA, 2010: 717-724.[27] Lawrence N. Probabilistic non-linear principal component analysis with Gaussian process latent variable models. Journal of Machine Learning Research, 2005, 6(3): 1783-1816.[28] Cai R, Xie F, Glymour C, et al. Triad Constraints for Learning Causal Structure of Latent Variables[C]\/\/Advances in Neural Information Processing Systems. 2019: 12863-12872.[29] Cai R, Qiao J, Zhang Z, et al. SELF: Structural Equational Likelihood Framework for Causal Discovery[C]\/\/Thirty-Second AAAI Conference on Artificial Intelligence. 2018.[30] Cai R, Zhang Zhen-Jie, Hao Zhi-Feng. SADA: A general framework to support robust causation discovery\/\/Proceedings of the 30th International Conference on Machine Learning. Atlanta, USA, 2013: 208-216.[31] Cai R, Zhang Z, Hao Z, et al. Sophisticated merging over random partitions: a scalable and robust causal discovery approach[J]. IEEE transactions on neural networks and learning systems, 2017, 29(8): 3623-3635.[32] Sch\u00f6lkopf B, Janzing D, Peters J, et al. On causal and anticausal learning[C]\/\/Proceedings of the 29th International Coference on International Conference on Machine Learning. 2012: 459-466.[33] Zhang K, Sch\u00f6lkopf B, Muandet K, et al. Domain adaptation under target and conditional shift[C]\/\/International Conference on Machine Learning. 2013: 819-827.[34] Zhang K, Gong M, Sch\u00f6lkopf B. Multi-source domain adaptation: A causal view[C]\/\/Twenty-ninth AAAI conference on artificial intelligence. 2015.[35] Cai R, Li Z, Wei P, et al. Learning disentangled semantic representation for domain adaptation[C]\/\/Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence. NIH Public Access, 2019.[36] Cai R, Li J, Zhang Z, et al. DACH: Domain Adaptation Without Domain Information[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020.(\u53c2\u8003\u6587\u732e\u53ef\u4e0a\u4e0b\u6ed1\u52a8\u67e5\u770b)<\/p>\n

\u9009\u81ea\u300a\u4e2d\u56fd\u4eba\u5de5\u667a\u80fd\u5b66\u4f1a\u901a\u8baf\u300b2020\u5e74  \u7b2c10\u5377  \u7b2c5\u671f  \u673a\u5668\u5b66\u4e60\u53ca\u5176\u5e94\u7528\u4e13\u9898<\/p>\n

\u56e0\u679c\u79d1\u5b66\u4e0eCausal AI\u8bfb\u4e66\u4f1a<\/p>\n


\u56e0\u679c\u63a8\u65ad\u4e0e\u673a\u5668\u5b66\u4e60\u9886\u57df\u7684\u7ed3\u5408\u5df2\u7ecf\u5438\u5f15\u4e86\u8d8a\u6765\u8d8a\u591a\u6765\u81ea\u5b66\u754c\u4e1a\u754c\u7684\u5173\u6ce8\uff0c\u4e3a\u6df1\u5165\u63a2\u8ba8\u3001\u666e\u53ca\u63a8\u5e7f\u56e0\u679c\u79d1\u5b66\u8bae\u9898\uff0c\u667a\u6e90\u793e\u533a\u643a\u624b\u96c6\u667a\u4ff1\u4e50\u90e8\u5c06\u4e3e\u529e\u7b2c\u4e8c\u671f\u300c\u56e0\u679c\u79d1\u5b66\u4e0eCausalAI\u8bfb\u4e66\u4f1a\u300d\u3002\u672c\u671f\u8bfb\u4e66\u4f1a\u7740\u529b\u4e8e\u5b9e\u64cd\u6027\u3001\u57fa\u7840\u6027\uff0c\u5c06\u5e26\u9886\u5927\u5bb6\u7cbe\u8bfb\u56e0\u679c\u79d1\u5b66\u65b9\u5411\u4e24\u672c\u975e\u5e38\u53d7\u5e7f\u6cdb\u8ba4\u53ef\u7684\u5165\u95e8\u6559\u6750\u3002
1. Pearl, Judea, Madelyn Glymour, and Nicholas P. Jewell. Causal inference in statistics: A primer. John Wiley & Sons, 2016.\uff08\u672c\u4e66\u4e2d\u8bd1\u7248\u300a\u7edf\u8ba1\u56e0\u679c\u63a8\u7406\u5165\u95e8\uff08\u7ffb\u8bd1\u7248\uff09\u300b\u5df2\u7531\u9ad8\u7b49\u6559\u80b2\u51fa\u7248\u793e\u51fa\u7248\uff09
2. Peters, Jonas, Dominik Janzing, and Bernhard Sch\u00f6lkopf. Elements of causal inference: foundations and learning algorithms. The MIT Press, 2017.
\u8bfb\u4e66\u4f1a\u6bcf\u5468\u5c06\u8fdb\u884c\u76f4\u64ad\u8ba8\u8bba\uff0c\u8fdb\u884c\u95ee\u9898\u4ea4\u6d41\u3001\u91cd\u70b9\u6982\u5ff5\u5206\u4eab\u3001\u9605\u8bfb\u6982\u89c8\u548c\u7f16\u7a0b\u5b9e\u8df5\u5185\u5bb9\u5206\u6790\u3002\u975e\u5e38\u9002\u5408\u6709\u673a\u5668\u5b66\u4e60\u80cc\u666f\uff0c\u5e0c\u671b\u6df1\u5165\u5b66\u4e60\u56e0\u679c\u79d1\u5b66\u57fa\u7840\u77e5\u8bc6\u548c\u91cd\u8981\u6a21\u578b\u65b9\u6cd5\uff0c\u5bfb\u6c42\u89e3\u51b3\u76f8\u5173\u7814\u7a76\u95ee\u9898\u7684\u670b\u53cb\u53c2\u52a0\u3002\u8be6\u60c5\u53c2\u89c1\uff1a\u8fde\u63a5\u7edf\u8ba1\u5b66\u3001\u673a\u5668\u5b66\u4e60\u4e0e\u81ea\u52a8\u63a8\u7406\u7684\u65b0\u5174\u4ea4\u53c9\u9886\u57df\u2014\u2014\u56e0\u679c\u79d1\u5b66\u8bfb\u4e66\u4f1a\u518d\u542f\u822a<\/a>\u3002

\u63a8\u8350\u9605\u8bfb
\u56e0\u679c\u8868\u5f81\u5b66\u4e60\u6700\u65b0\u7efc\u8ff0\uff1a\u8fde\u63a5\u56e0\u679c\u79d1\u5b66\u548c\u673a\u5668\u5b66\u4e60\u7684\u6865\u6881<\/a><\/a>\u56e0\u679c\u79d1\u5b66\u5165\u95e8\u8bfb\u4ec0\u4e48\u4e66\uff1fY. Bengio\u535a\u58eb\u5019\u9009\u4eba\u7684\u7814\u8bfb\u8def\u5f84\u63a8\u8350<\/a>\u56e0\u679c\u79d1\u5b66\uff1a\u8fde\u63a5\u7edf\u8ba1\u5b66\u3001\u673a\u5668\u5b66\u4e60\u4e0e\u81ea\u52a8\u63a8\u7406\u7684\u65b0\u5174\u4ea4\u53c9\u9886\u57df<\/a>
\u501f\u52a9\u56e0\u679c\u63a8\u65ad\uff0c\u66f4\u9c81\u68d2\u7684\u673a\u5668\u5b66\u4e60\u6765\u4e86\uff01<\/a>
\u5982\u4f55\u5728\u89c2\u6d4b\u6570\u636e\u4e0b\u8fdb\u884c\u56e0\u679c\u6548\u5e94\u8bc4\u4f30<\/a>
\u56e0\u679c\u9636\u68af\u4e0eDo-\u6f14\u7b97\uff1a\u600e\u6837\u5b8c\u7f8e\u5730\u8bc1\u660e\u5438\u70df\u81f4\u764c\uff1f<\/a>
\u52a0\u5165\u96c6\u667a\uff0c\u4e00\u8d77\u590d\u6742\uff01<\/a><\/p>\n

\u70b9\u51fb\u201c\u9605\u8bfb\u539f\u6587\u201d\uff0c\u62a5\u540d\u56e0\u679c\u79d1\u5b66\u4e0eCausal AI\u8bfb\u4e66\u4f1a\u7b2c\u4e8c\u5b63 <\/p><\/div>\n","protected":false},"excerpt":{"rendered":"

\u5bfc\u8bed \u672c\u6587\u4e3b\u8981\u4ece\u56e0\u679c\u63a8\u65ad\u65b9\u6cd5\u548c\u56e0\u679c\u6027\u5b66\u4e60\u8fd9\u4e24\u4e2a\u65b9\u9762\u4ecb\u7ecd\u548c\u5206\u6790\u4e86\u76ee\u524d\u56fd\u5185\u5916\u76f8\u5173\u9886\u57df\u7684\u7814\u7a76\u8fdb\u5c55\uff0c\u63a2\u8ba8\u4e86\u8fd9\u4e24\u65b9\u9762\u7814\u7a76\u7684\u53d1\u5c55\u65b9\u5411\u3002 \u96c6\u667a\u4ff1\u4e50\u90e8\u8054\u5408\u667a\u6e90\u793e\u533a\uff0c\u7ec4\u7ec7\u56e0\u679c\u79d1\u5b66\u7cfb\u5217\u8bfb\u4e66\u4f1a\u7b2c\u4e8c\u5b63\uff0c\u4ece\u57fa\u7840\u548c\u5b9e\u64cd\u89d2\u5ea6\u51fa\u53d1\uff0c\u7cbe\u8bfb\u4e24\u672c\u56e0\u679c\u79d1\u5b66\u65b9\u5411\u7684\u5165\u95e8\u6559\u6750\u3002\u672c\u6587\u4f5c\u8005\u8521\u745e\u521d\u8001\u5e08\u56e2\u961f\u4e5f\u4f1a\u5728\u8bfb\u4e66\u4f1a\u4e0a\u5c55\u5f00\u5206\u4eab\uff084\u6708\u5e95\uff09\uff0c\u8bfb\u4e66\u4f1a\u8be6\u60c5\u89c1…<\/p>\n","protected":false},"author":1,"featured_media":24380,"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\/24389"}],"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=24389"}],"version-history":[{"count":0,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/24389\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/media\/24380"}],"wp:attachment":[{"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=24389"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=24389"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=24389"},{"taxonomy":"special","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fspecial&post=24389"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}