{"id":40480,"date":"2025-09-28T22:05:16","date_gmt":"2025-09-28T14:05:16","guid":{"rendered":"https:\/\/17aitech.com\/?p=40480"},"modified":"2025-09-28T22:05:16","modified_gmt":"2025-09-28T14:05:16","slug":"%e6%b1%aa%e5%86%9b%e5%af%b9%e8%af%9d-rich-sutton%ef%bc%9a%e5%a4%a7%e6%a8%a1%e5%9e%8b%e5%9c%a8%e4%b8%80%e5%ae%9a%e7%a8%8b%e5%ba%a6%e4%b8%8a%e5%88%86%e6%95%a3%e4%ba%86%e6%88%91%e4%bb%ac%e5%af%b9","status":"publish","type":"post","link":"https:\/\/17aitech.com\/?p=40480","title":{"rendered":"\u6c6a\u519b\u5bf9\u8bdd Rich Sutton\uff1a\u5927\u6a21\u578b\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u6211\u4eec\u5bf9\u667a\u80fd\u7406\u89e3\u7684\u6ce8\u610f\u529b"},"content":{"rendered":"<p><\/p>\n<p>\u5728\u521a\u521a\u843d\u5e55\u7684 RL China 2025 \u5f00\u5e55\u5f0f\u4e0a\uff0c\u4f26\u6566\u5927\u5b66\u5b66\u9662\uff08UCL\uff09\u6c6a\u519b\u6559\u6388\u4e0e\u56fe\u7075\u5956\u5f97\u4e3b\u3001\u201c\u5f3a\u5316\u5b66\u4e60\u4e4b\u7236\u201d Richard Sutton \u5c55\u5f00\u4e86\u4e00\u573a\u8de8\u8d8a\u5730\u57df\u7684\u601d\u60f3\u5bf9\u8bdd\uff0c\u4ece\u5f3a\u5316\u5b66\u4e60\uff08RL\uff09\u7684\u5b66\u79d1\u6839\u57fa\u51fa\u53d1\uff0c\u5171\u63a2\u667a\u80fd\u7684\u672c\u8d28\u4e0e\u672a\u6765\u65b9\u5411\u3002<\/p>\n<p>\u6c6a\u519b\u6559\u6388\u6df1\u8015\u667a\u80fd\u4fe1\u606f\u7cfb\u7edf\u9886\u57df\u591a\u5e74\uff0c\u73b0\u4efb UCL \u8ba1\u7b97\u673a\u7cfb\u6559\u6388\uff0cTuring Fellow\uff0c\u662f\u534e\u4eba\u5f3a\u5316\u5b66\u4e60\u793e\u533a RL China \u7684\u8054\u5408\u53d1\u8d77\u4eba\u3002RL China \u662f\u7531\u5168\u7403\u534e\u4eba\u5b66\u8005\u4e0e\u5f3a\u5316\u5b66\u4e60\u76f8\u5173\u4ece\u4e1a\u8005\u5171\u540c\u53d1\u8d77\u7684\u975e\u76c8\u5229\u6027\u5b66\u672f\u4e0e\u6280\u672f\u4ea4\u6d41\u5e73\u53f0\uff0c\u81f4\u529b\u4e8e\u63a8\u52a8\u5f3a\u5316\u5b66\u4e60\u53ca\u51b3\u7b56\u667a\u80fd\u9886\u57df\u7684\u7814\u7a76\u3001\u5e94\u7528\u4e0e\u6559\u80b2\u3002<\/p>\n<p>Richard Sutton \u662f\u5f3a\u5316\u5b66\u4e60\u9886\u57df\u7684\u5960\u57fa\u8005\uff0c\u8fd9\u4f4d\u7f8e\u56fd\u8ba1\u7b97\u673a\u79d1\u5b66\u5bb6\u73b0\u4efb\u963f\u5c14\u4f2f\u5854\u5927\u5b66\u8ba1\u7b97\u673a\u5b66\u6559\u6388\uff0c\u56e0\u63d0\u51fa\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u3001\u7b56\u7565\u68af\u5ea6\u65b9\u6cd5\u7b49\u6838\u5fc3\u7406\u8bba\uff0c\u88ab\u516c\u8ba4\u4e3a \u201c\u73b0\u4ee3\u5f3a\u5316\u5b66\u4e60\u4e4b\u7236\u201d\u30022024 \u5e74\uff0c\u4ed6\u65a9\u83b7\u8ba1\u7b97\u673a\u9886\u57df\u6700\u9ad8\u8363\u8a89\u2014\u2014\u56fe\u7075\u5956\uff0c\u8fd9\u4e00\u5956\u9879\u4e0d\u4ec5\u662f\u5bf9\u5176\u5728\u7b97\u6cd5\u521b\u65b0\u4e0e\u57fa\u7840\u7814\u7a76\u4e2d\u7a81\u51fa\u8d21\u732e\u7684\u8ba4\u53ef\uff0c\u66f4\u6807\u5fd7\u7740\u5f3a\u5316\u5b66\u4e60\u4f5c\u4e3a AI \u6838\u5fc3\u5206\u652f\u7684\u4ef7\u503c\u4e0e\u5f71\u54cd\u529b\uff0c\u4ee5\u53ca\u6574\u4e2a RL \u7814\u7a76\u793e\u533a\u4e3a\u4eba\u5de5\u667a\u80fd\u53d1\u5c55\u5960\u5b9a\u7684\u57fa\u7840\u3002<\/p>\n<p>Richard \u5728\u5bf9\u8bdd\u4e2d\u5766\u8a00\uff1a\u5f53\u524d\u4eba\u5de5\u667a\u80fd\u884c\u4e1a\u7684\u5feb\u901f\u6269\u5f20\u867d\u6539\u53d8\u4e86\u4eba\u4eec\u4ece\u4e8b\u7684\u5de5\u4f5c\u4e0e\u751f\u6d3b\u611f\u77e5\uff0c\u5374\u4e5f\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u5bf9\u57fa\u7840\u79d1\u5b66\u7684\u5173\u6ce8\uff0c\u8fd9\u5bf9\u79d1\u5b66\u53d1\u5c55\u6765\u8bf4\u5e76\u975e\u771f\u6b63\u6709\u76ca\u3002\u6211\u4eec\u5fc5\u987b\u5bf9\u6b64\u8fdb\u884c\u53cd\u51fb\uff0c\u52aa\u529b\u5c06\u76ee\u5149\u6295\u5411\u7406\u89e3\u601d\u7ef4\u5982\u4f55\u8fd0\u4f5c\u3001\u667a\u80fd\u5982\u4f55\u8fd0\u4f5c\u8fd9\u4e00\u91cd\u5927\u76ee\u6807\u3002<\/p>\n<p>\u5728\u4ed6\u770b\u6765\uff0c\u5927\u6a21\u578b\u5e76\u975e\u771f\u6b63\u7406\u89e3\u667a\u80fd\u7684\u9014\u5f84\u3002LLM \u786e\u5b9e\u5f88\u6709\u7528\uff0c\u4f46\u5b83\u4eec\u5728\u8fd0\u884c\u8fc7\u7a0b\u4e2d\u662f\u6ca1\u6709\u5b66\u4e60\u80fd\u529b\u7684\uff0c\u4e00\u65e6\u6a21\u578b\u88ab\u8bad\u7ec3\u5b8c\u6210\uff0c\u5b83\u4eec\u5c31\u4e0d\u518d\u5b66\u4e60\u4e86\u3002\u4e8b\u5b9e\u4e0a\uff0c\u5b83\u4eec\u4e5f\u65e0\u6cd5\u5b66\u4e60\uff0c\u56e0\u4e3a\u5b83\u4eec\u6ca1\u6709\u76ee\u6807\uff0c\u6ca1\u6709\u5956\u52b1\u3002\u5b83\u4eec\u53ea\u662f\u5728\u6a21\u4eff\u4eba\u7c7b\uff0c\u800c\u4eba\u7c7b\u662f\u6709\u76ee\u6807\u7684\uff0c\u4f46 LLM \u672c\u8eab\u6ca1\u6709\u76ee\u6807\uff0c\u4e5f\u65e0\u6cd5\u5224\u65ad\u81ea\u5df1\u8bf4\u7684\u8bdd\u662f\u5bf9\u662f\u9519\u3002<\/p>\n<p>\u4eba\u7c7b\u8bed\u5883\u4e2d\u7684\u76ee\u6807\u4e0e\u610f\u56fe\u4e00\u822c\u88ab\u7406\u89e3\u4e3a\uff1a\u5bf9\u4e00\u79cd\u540d\u4e3a \u201c\u5956\u52b1\u201d \u7684\u6807\u91cf\u4fe1\u53f7\u7684\u7d2f\u79ef\u548c\u7684\u671f\u671b\u503c\u8fdb\u884c\u6700\u5927\u5316\uff0c\u8fd9\u662f\u5bf9\u667a\u80fd\u95ee\u9898\u4e00\u9879\u6781\u5177\u4ef7\u503c\u7684\u7b80\u5316\u4e0e\u63d0\u70bc\u3002\u8981\u77e5\u9053\uff0c\u667a\u80fd\u5c31\u662f\u4e0e\u4e16\u754c\u4ea4\u4e92\u4ee5\u8fbe\u6210\u76ee\u6807\uff0c\u800c\u4f60\u9700\u8981\u67d0\u79cd\u65b9\u5f0f\u6765\u5b9a\u4e49\u8fd9\u4e2a\u76ee\u6807\u3002\u6240\u4ee5\uff0c\u5728\u5f3a\u5316\u5b66\u4e60\u7684\u6846\u67b6\u4e0b\uff0cRichard \u5efa\u8bae\uff1a<strong>\u6240\u6709\u76ee\u6807\u90fd\u53ef\u4ee5\u88ab\u89e3\u8bfb\u4e3a \u201c\u5956\u52b1\u201d\uff0c\u4e00\u79cd\u5355\u4e00\u7684\u6807\u91cf\u6570\u503c\u3002<\/strong><\/p>\n<p>\u6b63\u56e0\u5982\u6b64\uff0c Richard \u975e\u5e38\u91cd\u89c6\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\uff0c\u4ed6\u8ba4\u4e3a\uff0c\u4ece\u67d0\u79cd\u610f\u4e49\u4e0a\u6765\u8bf4\uff0c<strong>RL \u7684\u6838\u5fc3\u5728\u4e8e\uff0c\u5b83\u6700\u5f3a\u5927\u7684\u601d\u7ef4\u65b9\u5f0f\u6765\u81ea\u4e8e\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\u7684\u7406\u5ff5<\/strong>\u3002\u800c\u4e14\u65e2\u5e0c\u671b\u76f4\u63a5\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\uff0c\u4e5f\u5e0c\u671b\u95f4\u63a5\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\uff0c\u8ba9\u81ea\u5df1\u7684\u7ecf\u9a8c\u80fd\u5f62\u6210\u4e00\u4e2a\u4e16\u754c\u6a21\u578b\uff0c\u7136\u540e\u5229\u7528\u8fd9\u4e2a\u6a21\u578b\u6765\u89c4\u5212\u884c\u4e3a\u4e0a\u7684\u6539\u8fdb\u3002\u8fd9\u79cd\u89c4\u5212\u5c31\u7c7b\u4f3c\u4e8e\u63a8\u7406\uff0c\u4eba\u4eec\u4ec5\u51ed\u5bf9\u4e16\u754c\u7684\u8ba4\u77e5\u5c31\u80fd\u60f3\u6e05\u695a\u8be5\u505a\u4ec0\u4e48\u3002<\/p>\n<p>\u800c\u4e14\u667a\u80fd\u7684\u57fa\u672c\u539f\u5219\u6709\u9650\uff0c<strong>\u68af\u5ea6\u4e0b\u964d\u5728\u5176\u4e2d\u81f3\u5173\u91cd\u8981<\/strong>\u3002\u4f46\u884c\u4e1a\u719f\u7ec3\u540e\u5f88\u5bb9\u6613\u9677\u5165\u53ea\u7528\u68af\u5ea6\u4e0b\u964d\uff0c\u800c\u5ffd\u89c6\u5176\u4ed6\u53ef\u80fd\u6027\u7684\u5c40\u9650\u4e2d\u3002<strong>\u5b9e\u9645\u4e0a\u68af\u5ea6\u4e0b\u964d\u3001\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u3001\u641c\u7d22\u90fd\u662f\u667a\u80fd\u7684\u57fa\u672c\u539f\u5219\u7684\u5176\u4e2d\u4e4b\u4e00\uff0c\u5176\u4e2d\u641c\u7d22\u5c24\u4e3a\u5173\u952e<\/strong>\u3002\u968f\u673a\u6743\u91cd\u662f\u641c\u7d22\u7684\u6838\u5fc3\u8981\u7d20\uff0c\u5b83\u80fd\u8ba9\u6a21\u578b\u63a2\u7d22\u66f4\u591a\u53ef\u80fd\u6027\u3002\u56e0\u6b64\uff0c\u68af\u5ea6\u4e0b\u964d\u7b97\u6cd5\u4e2d\u9700\u8981\u878d\u5165\u8fd9\u79cd\u641c\u7d22\u6210\u5206\uff0c\u628a\u641c\u7d22\u548c\u68af\u5ea6\u4e0b\u964d\u7ed3\u5408\u8d77\u6765\uff0c\u5176\u5a01\u529b\u4f1a\u8fdc\u5927\u4e8e\u68af\u5ea6\u4e0b\u964d\u5355\u72ec\u4f7f\u7528\u3002<\/p>\n<p>\u4ee5\u4e0b\u662f\u6b64\u6b21\u5bf9\u8bdd\u7684\u7cbe\u5f69\u5206\u4eab\uff0cAI \u79d1\u6280\u8bc4\u8bba\u8fdb\u884c\u4e86\u4e0d\u6539\u539f\u610f\u7684\u7f16\u8f91\u6574\u7406\uff1a<\/p>\n<p style=\"text-align:center\" class=\"heading-2 ace-line old-record-id-LKuydYhLnobOuxxxZxDc1pDMnWx\"><span style=\"font-size: 20px\"><strong>01 \u5bf9\u667a\u80fd\u7684\u601d\u8003<\/strong><\/span><\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u53ef\u4ee5\u7b80\u5355\u4ecb\u7ecd\u4e00\u4e0b\u4f60 2023 \u5e74\u521b\u5efa Openmind Research Institute\uff08\u5f00\u653e\u601d\u7ef4\u7814\u7a76\u6240\uff09\u7684\u80cc\u666f\u5417\uff1f\u4ee5\u53ca\u4f60\u60f3\u8981\u5b9e\u73b0\u7684\u4f7f\u547d\u662f\u4ec0\u4e48\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1aOpenmind \u00a0Research Institute \u7684\u4f7f\u547d\u662f<strong>\u7406\u89e3\u667a\u80fd\uff0c\u5e76\u4e0e\u4e16\u754c\u5206\u4eab\u8fd9\u79cd\u7406\u89e3<\/strong>\u3002\u5c1d\u8bd5\u4e0e\u4e16\u754c\u5206\u4eab\u7684\u539f\u56e0\u662f\uff0c\u4eba\u5de5\u667a\u80fd\u5df2\u7ecf\u53d8\u5f97\u6709\u70b9\u5730\u7f18\u653f\u6cbb\u5316\uff0c\u5404\u4e2a\u56fd\u5bb6\u90fd\u5728\u8bd5\u56fe\u4e3b\u5bfc\u8fd9\u4e2a\u9886\u57df\u3002\u5f53\u7136\uff0c\u8bb8\u591a\u4e0d\u540c\u7684\u653f\u6cbb\u9635\u8425\u4e4b\u95f4\u5b58\u5728\u7ade\u4e89\uff0c\u5c24\u5176\u662f\u4e2d\u7f8e\u3002\u4efb\u4f55\u4e00\u65b9\u90fd\u4e0d\u5e94\u8be5\u5360\u636e\u4e3b\u5bfc\u5730\u4f4d\u3002<\/p>\n<p><strong>\u7406\u89e3\u667a\u80fd\u662f\u5bf9\u4eba\u7c7b\u601d\u7ef4\u7406\u89e3\u7684\u6839\u672c\u6027\u63a2\u7d22<\/strong>\uff0c\u6bcf\u4e2a\u4eba\u90fd\u5e94\u8be5\u53c2\u4e0e\u5176\u4e2d\u3002\u8fd9\u4e3b\u8981\u662f\u4e00\u9879\u79d1\u5b66\u63a2\u7d22\uff0c\u800c\u4e0d\u662f\u4e3a\u4e86\u8ffd\u6c42\u4ea7\u4e1a\u53d1\u5c55\u3002\u5927\u5bb6\u90fd\u5e94\u8be5\u5c1d\u8bd5\u9762\u5bf9\u601d\u7ef4\u7684\u6839\u672c\u95ee\u9898\uff0c\u4ee5\u53ca\u5982\u4f55\u7406\u89e3\u5b83\u3001\u5982\u4f55\u8ba9\u5b83\u66f4\u597d\u5730\u8fd0\u4f5c\uff0c\u8fd9\u6837\u6211\u4eec\u624d\u80fd\u66f4\u597d\u5730\u8fd0\u4f5c\uff0c\u624d\u80fd\u7406\u89e3\u6211\u4eec\u662f\u4ec0\u4e48\uff0c\u4ee5\u53ca\u6211\u4eec\u5728\u5b87\u5b99\u4e2d\u626e\u6f14\u4ec0\u4e48\u89d2\u8272\u3002<\/p>\n<p>Openmind Research Institute \u80af\u5b9a\u662f\u4ece\u5c0f\u4e8b\u505a\u8d77\uff0c\u5b83\u662f\u4e00\u4e2a\u5b9e\u9a8c\u5ba4\uff0c\u4e2d\u5fc3\u5728\u52a0\u62ff\u5927\u963f\u5c14\u4f2f\u5854\u7701\uff0c\u4f46\u4e4b\u540e\u8fd8\u4f1a\u542f\u52a8\u4e00\u4e2a\u4ee5\u65b0\u52a0\u5761\u4e3a\u4e2d\u5fc3\u7684\u65b0\u5b9e\u9a8c\u5ba4\u3002\u8fd9\u6837\u4e00\u6765\uff0c\u7814\u7a76\u8303\u56f4\u5c31\u904d\u5e03\u5168\u7403\u4e86\uff0c\u6bcf\u4e2a\u5b9e\u9a8c\u5ba4\u90fd\u4f1a\u6709\u4e00\u5b9a\u6570\u91cf\u7684\u7814\u7a76\u5458\uff0c\u901a\u5e38\u662f\u521a\u83b7\u5f97\u535a\u58eb\u5b66\u4f4d\u4f46\u60f3\u8981\u4ece\u4e8b\u57fa\u7840\u7814\u7a76\u548c\u4eba\u5de5\u667a\u80fd\u7684\u5e74\u8f7b\u4eba\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u8fd9\u662f\u8981\u5fc5\u987b\u505a\u7684\uff0c\u800c\u4e14\u975e\u5e38\u53ca\u65f6\uff0c\u56e0\u4e3a\u73b0\u5728\u6709\u4e9b\u7814\u7a76\u793e\u533a\u751a\u81f3\u4e0d\u53d1\u8868\u8bba\u6587\u3002\u8fd9\u8c08\u4e0d\u4e0a\u7edd\u5bf9\u7684\u597d\u574f\uff0c\u8fd9\u4e9b\u7b97\u6cd5\u672c\u8d28\u4e0a\u662f\u7279\u5b9a\u65f6\u671f\u7684\u7814\u7a76\u6210\u679c\uff0c\u95ee\u9898\u5728\u4e8e\uff0c\u6211\u4eec\u4f3c\u4e4e\u5df2\u7ecf\u4e22\u5931\u4e86\u5f53\u521d\u9a71\u52a8\u5176\u8bde\u751f\u7684\u53d1\u5c55\u52bf\u5934\uff0c\u8fd9\u79cd\u72b6\u6001\u5bf9\u6574\u4e2a\u7814\u7a76\u5b66\u754c\u6765\u8bf4\u662f\u6781\u4e0d\u5065\u5eb7\u7684\u3002\uff08Openmind Research Institute\uff09\u8fd9\u4e2a\u60f3\u6cd5\u8ba9\u4e00\u5207\u90fd\u53d8\u5f97\u5f00\u653e\u548c\u72ec\u7acb\uff0c\u8ba9\u79d1\u5b66\u56de\u5f52\u5230\u7eaf\u7cb9\u7684\u79d1\u5b66\u597d\u5947\u5fc3\uff0c\u8fd9\u5f88\u5173\u952e\u3002\u6211\u5f88\u9ad8\u5174\u73b0\u5728\u4e00\u4e9b\u57fa\u7840\u6a21\u578b\u5f00\u59cb\u5f00\u6e90\u6216\u5f00\u653e\u53c2\u6570\uff0c\u60c5\u51b5\u53d8\u5f97\u8d8a\u6765\u8d8a\u597d\u4e86\uff0c\u5e0c\u671b Openmind \u00a0Research Institute \u80fd\u5728\u4e2d\u56fd\u8bbe\u7acb\u5206\u652f\u673a\u6784\uff0c\u9f13\u52b1\u5e74\u8f7b\u4eba\uff0c\u8ba9\u5e74\u8f7b\u7684\u7814\u7a76\u4eba\u5458\u56de\u5f52\u771f\u6b63\u7684\u79d1\u5b66\u8ffd\u6c42\uff0c\u800c\u4e0d\u53ea\u662f\u8ffd\u6c42\u4ea7\u4e1a\u8fdb\u6b65\u3002<\/p>\n<p>\u6211\u5728 RL China \u4e3e\u529e\u8fc7\u4e00\u4e9b\u4f1a\u8bae\u7814\u8ba8\u4f1a\uff0c\u5df2\u7ecf\u4e3e\u529e\u4e86\u597d\u51e0\u5e74\uff0c\u8fd8\u6709\u8fd9\u6b21\u590f\u4ee4\u8425\u3002\u5927\u591a\u6570\u53c2\u4f1a\u8005\u90fd\u662f\u5e74\u8f7b\u7684\u7814\u7a76\u5458\uff0c\u5c24\u5176\u662f\u5728 RL \u9886\u57df\uff0c\u4ee5\u53ca\u4ece\u673a\u5668\u4eba\u6280\u672f\u5230\u591a\u667a\u80fd\u4f53\uff08Multi-Agent\uff09\u7cfb\u7edf\uff0c\u518d\u5230\u6838\u5fc3\u9886\u57df\u3002\u4f60\u5bf9\u5e74\u8f7b\u7684\u7814\u7a76\u4eba\u5458\u6709\u4ec0\u4e48\u60f3\u8bf4\u7684\u5417\uff1f\u6216\u8005\u8ba4\u4e3a\u4ed6\u4eec\u5e94\u8be5\u601d\u8003\u7684\u6839\u672c\u95ee\u9898\u662f\u4ec0\u4e48\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u6709\u4e00\u70b9\u5f88\u5173\u952e\uff1a\u8981<strong>\u7740\u773c\u4e8e\u957f\u8fdc\u76ee\u6807<\/strong>\u3002<\/p>\n<p>\u4eba\u5de5\u667a\u80fd\u4e0d\u662f\u4eca\u5e74\u6216\u660e\u5e74\u5c31\u80fd\u5b9e\u73b0\u7684\uff0c\u5b83\u53ef\u80fd\u4f1a\u5728\u8fd9\u4e2a\u5341\u5e74\u6216\u4e0b\u4e2a\u5341\u5e74\u5b8c\u6210\uff0c\u6240\u4ee5\u8fd9\u662f\u4e00\u4e2a\u957f\u8fdc\u76ee\u6807\uff0c\u5b83\u5c31\u50cf\u4e00\u573a\u9a6c\u62c9\u677e\uff0c\u800c\u4e0d\u662f\u77ed\u8dd1\u6bd4\u8d5b\u3002\u5c24\u5176\u5728\u4eba\u5de5\u667a\u80fd\u7684 RL \u9886\u57df\u4e2d\uff0c\u60c5\u51b5\u6bd4\u8f83\u7279\u6b8a\uff1a\u4e00\u65b9\u9762\uff0c\u6211\u4eec\u9762\u4e34\u7740\u4e00\u4e2a\u57fa\u7840\u6027\u7684\u79d1\u5b66\u96be\u9898\uff1b\u4f46\u53e6\u4e00\u65b9\u9762\uff0c\u8fd9\u4e2a\u9886\u57df\u53c8\u5df2\u7ecf\u5f62\u6210\u4e86\u5e9e\u5927\u7684\u4ea7\u4e1a\uff0c\u6709\u5de8\u989d\u8d44\u91d1\u6e90\u6e90\u4e0d\u65ad\u5730\u6295\u5165\u5230\u5404\u7c7b\u5e94\u7528\u4e2d\u3002\u5355\u770b\u4eba\u5de5\u667a\u80fd\u8fd9\u4e2a\u884c\u4e1a\u7684\u786e\u5341\u5206\u5e9e\u5927\uff0c\u5b83\u4e5f\u786e\u5b9e\u6539\u53d8\u4e86\u4eba\u4eec\u7684\u611f\u53d7\u4ee5\u53ca\u4eba\u4eec\u53ef\u4ee5\u4ece\u4e8b\u7684\u5de5\u4f5c\uff0c\u4f46<strong>\u540c\u65f6\u4e5f\u5728\u67d0\u79cd\u7a0b\u5ea6\u4e0a\u51cf\u5c11\u4e86\u5bf9\u57fa\u7840\u79d1\u5b66\u7684\u5173\u6ce8<\/strong>\u3002<\/p>\n<p>\u5b83\u66f4\u50cf\u662f\u4e00\u573a\u5feb\u901f\u7684\u6e38\u620f\uff0c\u4f60\u53ef\u4ee5\u5f88\u5feb\u5c31\u8d5a\u70b9\u94b1\u9000\u51fa\u3002\u8fd9\u5bf9\u79d1\u5b66\u53d1\u5c55\u6765\u8bf4\u5e76\u975e\u771f\u6b63\u6709\u76ca\u3002<\/p>\n<p>\u4ece\u5b8f\u89c2\u89d2\u5ea6\u6765\u770b\uff0c\u79d1\u5b66\u53d6\u5f97\u5982\u6b64\u5927\u7684\u6210\u529f\u662f\u4ef6\u597d\u4e8b\uff0c\u8fd9\u4e3a\u8be5\u9886\u57df\u5e26\u6765\u4e86\u5927\u91cf\u8d44\u91d1\u3002\u4f46\u4f5c\u4e3a\u4e00\u95e8\u79d1\u5b66\uff0c\u5b83\u5728\u67d0\u79cd\u7a0b\u5ea6\u4e0a\u6539\u53d8\u4e86\u8fd9\u4e2a\u9886\u57df\u7684\u6027\u8d28\uff0c\u6211\u4eec\u5fc5\u987b\u5bf9\u6b64\u8fdb\u884c\u53cd\u51fb\uff0c\u52aa\u529b\u5c06\u76ee\u5149\u6295\u5411\u7406\u89e3<strong>\u601d\u7ef4\u5982\u4f55\u8fd0\u4f5c\u3001\u667a\u80fd\u5982\u4f55\u8fd0\u4f5c<\/strong>\u8fd9\u4e00\u91cd\u5927\u76ee\u6807\uff0c\u5e76\u4e0d\u65ad\u89e3\u51b3\u95ee\u9898\u3002\u65e2\u8981\u7740\u624b\u89e3\u51b3\u90a3\u4e9b\u6211\u4eec\u5df2\u7ecf\u77e5\u9053\u8be5\u600e\u4e48\u505a\u7684\u95ee\u9898\uff0c\u66f4\u8981\u6301\u7eed\u5173\u6ce8\u90a3\u4e9b\u6211\u4eec\u8fd8\u4e0d\u77e5\u9053\u8be5\u5982\u4f55\u89e3\u51b3\u7684\u95ee\u9898\u3002\u5de5\u4e1a\u754c\u503e\u5411\u4e8e\u5173\u6ce8\u6211\u4eec\u5df2\u77e5\u7684\u3001\u6211\u4eec\u4eca\u5929\u80fd\u505a\u4ec0\u4e48\uff1b\u800c<strong>\u79d1\u5b66\u754c\u5219\u5173\u6ce8\u6211\u4eec\u5c1a\u4e0d\u77e5\u9053\u5982\u4f55\u505a\u7684\u4e8b\u60c5<\/strong>\uff0c\u662f\u4ec0\u4e48\u963b\u788d\u4e86\u6211\u4eec\u6df1\u5165\u7406\u89e3\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u6ca1\u9519\uff0c\u6211\u8ba4\u4e3a\u5728\u4e2d\u56fd\u4e5f\u6709\u7c7b\u4f3c\u7684\u60c5\u51b5\u3002\u4ea7\u4e1a\u65b9\u7684\u53c2\u4e0e\u5728\u67d0\u4e9b\u65b9\u9762\u786e\u5b9e\u975e\u5e38\u6709\u76ca\uff0c\u4f46\u53e6\u4e00\u65b9\u9762\uff0c\u5b83\u5728\u67d0\u79cd\u7a0b\u5ea6\u4e0a\u4f1a\u63a8\u52a8\u751a\u81f3\u662f\u8feb\u4f7f\u7814\u7a76 \uff08\u5c24\u5176\u662f\u5b66\u672f\u7814\u7a76\uff09\u805a\u7126\u4e8e\u4e00\u4e9b\u77ed\u671f\u95ee\u9898\uff0c\u8fd9\u672a\u5fc5\u662f\u597d\u4e8b\u3002\u6bd4\u5982\u5927\u6a21\u578b\uff08LLM\uff09\uff0c\u9700\u8981\u8fdb\u884c\u9884\u8bad\u7ec3\uff0c\u8fd8\u9700\u8981\u5927\u91cf\u7684\u8ba1\u7b97\u8d44\u6e90\u3002\u8fd9\u786e\u5b9e\u662f\u4e00\u90e8\u5206\u4eba\u80fd\u505a\u7684\u4e8b\uff0c\u4f46\u5bf9\u4e8e\u6574\u4e2a\u7814\u7a76\u754c\u6765\u8bf4\uff0c\u53ef\u80fd\u65e0\u6cd5\u90fd\u6295\u5165\u5176\u4e2d\u3002\u6211\u4eec\u771f\u6b63\u5e94\u8be5\u5173\u6ce8\u7684\u662f<strong>\u8d85\u8d8a Transformer \u7684\u4e1c\u897f\uff0c\u8d85\u8d8a\u4f60\u4e4b\u524d\u63d0\u51fa\u5e76\u7814\u7a76\u8fc7\u7684\u91cd\u8981\u5b66\u4e60\u9886\u57df\u3002<\/strong><\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u8ba9\u6211\u518d\u6b21\u5f3a\u8c03\u4e00\u4e0b\u5173\u4e8e LLM \u7684\u8ba8\u8bba\uff0cLLM \u5728\u4eba\u5de5\u667a\u80fd\u4ea7\u4e1a\u4e2d\u5360\u636e\u4e3b\u5bfc\u5730\u4f4d\uff0c\u4f46\u6211\u786e\u5b9e\u89c9\u5f97\u5b83\u5bf9\u79d1\u5b66\u9020\u6210\u4e86\u5de8\u5927\u7684\u5e72\u6270\u3002<\/p>\n<p>\u6240\u4ee5\u6211\u60f3\u5bf9\u5404\u4f4d\u8bf4\u7684\u662f\uff0c\u867d\u7136 RL \u53ef\u4ee5\u4f5c\u4e3a LLM \u7684\u4e00\u90e8\u5206\u2014\u2014\u8fd9\u6ca1\u95ee\u9898\uff0c\u800c\u4e14 LLM \u786e\u5b9e\u662f\u4e00\u79cd\u4e0e\u4e92\u8054\u7f51\u6240\u6709\u77e5\u8bc6\u8fdb\u884c\u4ea4\u4e92\u7684\u6709\u6548\u65b9\u5f0f\uff0c\u5b83\u6709\u5f88\u591a\u4f18\u70b9\u2014\u2014<strong>\u4f46\u5b83\u5728\u5f88\u5927\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u6211\u4eec\u5bf9\u7406\u89e3\u667a\u80fd\u95ee\u9898\u7684\u6ce8\u610f\u529b\u3002<\/strong><\/p>\n<p>\u6211\u4eec\u5fc5\u987b\u628a\u5b83\u653e\u5728\u4e00\u8fb9\uff0c\u4e0d\u8981\u8d70\u8fd9\u6761\u8def\u3002\u5b66\u751f\u4eec\u5bf9 LLM \u7684\u5173\u6ce8\u548c\u70ed\u60c5\uff0c\u5f88\u96be\u4e0d\u60f3\u5728\u6b64\u57fa\u7840\u4e0a\u7ee7\u7eed\u7814\u7a76\uff0c\u4f46\u5b83<strong>\uff08\u5927\u6a21\u578b\uff09\u5e76\u975e\u771f\u6b63\u7406\u89e3\u667a\u80fd\u7684\u9014\u5f84<\/strong>\u3002\u867d\u7136 LLM \u786e\u5b9e\u5f88\u6709\u7528\uff0c\u4f46\u5b83\u4eec\u5728\u8fd0\u884c\u8fc7\u7a0b\u4e2d\u662f\u6ca1\u6709\u5b66\u4e60\u80fd\u529b\u7684\uff0c\u673a\u5668\u5b66\u4e60\u786e\u5b9e\u88ab\u7528\u4e8e\u521b\u5efa\u5b83\u4eec\uff0c\u4f46\u4e00\u65e6\u6a21\u578b\u88ab\u8bad\u7ec3\u5b8c\u6210\uff0c\u5b83\u4eec\u5c31\u4e0d\u518d\u5b66\u4e60\u4e86\u3002<\/p>\n<p>\u4e8b\u5b9e\u4e0a\uff0c<strong>\u5b83\u4eec\u4e5f\u65e0\u6cd5\u5b66\u4e60\uff0c\u56e0\u4e3a\u5b83\u4eec\u6ca1\u6709\u76ee\u6807\uff0c\u6ca1\u6709\u5956\u52b1\u3002\u5b83\u4eec\u53ea\u662f\u5728\u6a21\u4eff\u4eba\u7c7b\uff0c\u800c\u4eba\u7c7b\u662f\u6709\u76ee\u6807\u7684\uff0c\u4f46 LLM \u672c\u8eab\u6ca1\u6709\u76ee\u6807\uff0c\u4e5f\u65e0\u6cd5\u5224\u65ad\u81ea\u5df1\u8bf4\u7684\u8bdd\u662f\u5bf9\u662f\u9519\u3002<\/strong><\/p>\n<p>\u6211\u6709\u4e00\u53e5\u7528\u4e8e\u4eba\u5de5\u667a\u80fd\u7814\u7a76\u7684\u53e3\u53f7\uff0c\u5982\u679c\u5728\u7f51\u4e0a\u641c\u7d22\u201cRich&#8217;s slogan\u201d\u4e4b\u7c7b\u7684\u8bcd\uff0c\u5c31\u53ef\u4ee5\u627e\u5230\u3002\u5176\u4e2d\u4e00\u4e2a\u6838\u5fc3\u53e3\u53f7\u5c31\u662f\uff1a<strong>\u4f60\u4e0d\u5e94\u8be5\u8981\u6c42\u4f60\u7684\u4eba\u5de5\u667a\u80fd\u77e5\u9053\u4e00\u4e9b\u5b83\u81ea\u5df1\u90fd\u65e0\u6cd5\u5224\u65ad\u6b63\u786e\u7684\u4e8b\u60c5\u3002Agent\uff08\u667a\u80fd\u4f53\uff09\u5fc5\u987b\u80fd\u591f\u9a8c\u8bc1\u5b83\u7684\u77e5\u8bc6\uff0cLLM \u65e0\u6cd5\u9a8c\u8bc1\u5b83\u4eec\u7684\u4efb\u4f55\u77e5\u8bc6\uff0c\u9664\u4e86\u4eba\u7c7b\u4f1a\u8bf4\u5b83\u6b63\u786e\u4e4b\u5916\uff0c\u4f46\u8fd9\u79cd\u6b63\u786e\u6ca1\u6709\u4efb\u4f55\u610f\u4e49\u3002<\/strong><\/p>\n<p>\u90a3\u4e48\u8be5\u5982\u4f55\u9a8c\u8bc1\u4f60\u7684\u77e5\u8bc6\u5462\uff1f\u53ef\u80fd\u6709\u4e00\u79cd\u884c\u4e3a\u65b9\u5f0f\uff1a\u5982\u679c\u5b83\u6ca1\u6709\u5f97\u5230\u5956\u52b1\uff0c\u4f60\u5c31\u77e5\u9053\u8fd9\u4e0d\u662f\u4e00\u4e2a\u597d\u7684\u884c\u4e3a\u65b9\u5f0f\uff0c\u8fd9\u5c31\u7ed9\u4e86\u4f60\u5956\u52b1\uff0c\u7ed9\u4e86\u4f60\u4e00\u4e2a\u76ee\u6807\uff0c\u6216\u8005\u4f60\u53ef\u80fd\u6709\u4e00\u4e2a\u4e16\u754c\u7684\u9884\u6d4b\u6a21\u578b\uff0c\u4f60\u53ef\u80fd\u4f1a\u9884\u6d4b\u67d0\u4e8b\u4f1a\u53d1\u751f\uff0c\u7136\u540e\u4ece\u7ecf\u9a8c\u4e2d\u770b\u5230\u5b83\u786e\u5b9e\u53d1\u751f\u4e86\u3002\u4f46 LLM \u4e0d\u4f1a\u8fd9\u6837\u505a\uff0c\u5b83\u4eec\u53ef\u80fd\u4f1a\u8bf4\u6211\u9884\u6d4b\u4e86\u67d0\u4e8b\uff0c<strong>\u4f46\u5927\u6a21\u578b\u65e0\u6cd5\u5c06\u5176\u4e0e\u5b9e\u9645\u53d1\u751f\u7684\u4e8b\u60c5\u8fdb\u884c\u6bd4\u8f83\uff0c\u56e0\u4e3a\u5927\u6a21\u578b\u6ca1\u6709\u6301\u7eed\u7684\u7ecf\u9a8c\u6d41\u3002<\/strong><\/p>\n<p>\u6240\u4ee5\u6211\u975e\u5e38\u91cd\u89c6\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\uff0c\u4ece\u67d0\u79cd\u610f\u4e49\u4e0a\u6765\u8bf4\uff0c<strong>RL \u7684\u6838\u5fc3\u5728\u4e8e\uff0c\u5b83\u6700\u5f3a\u5927\u7684\u601d\u7ef4\u65b9\u5f0f\u6765\u81ea\u4e8e\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\u7684\u7406\u5ff5\uff0c\u8fd9\u79cd\u7406\u5ff5\u5b9e\u9645\u4e0a\u53ef\u4ee5\u8ffd\u6eaf\u5230\u4eba\u5de5\u667a\u80fd\u8bde\u751f\u4e4b\u521d<\/strong>\u3002<\/p>\n<p>\u827e\u4f26\u00b7\u56fe\u7075\u6709\u4e00\u7bc7\u4e66\u9762\u6f14\u8bb2\u63d0\u5230\uff1a\u6211\u4eec\u60f3\u8981\u7684\u662f\u4e00\u4e2a\u80fd\u591f\u57281947\u5e74\u3001\u751a\u81f3\u5728\u4eba\u5de5\u667a\u80fd\u51fa\u73b0\u4e4b\u524d\uff0c\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\u7684\u673a\u5668\u3002\u6211\u4eec\u73b0\u5728\u4f9d\u7136\u6e34\u671b\u8fd9\u6837\u7684\u673a\u5668\uff0c\u8fd9\u4ecd\u7136\u662f\u6838\u5fc3\uff0c\u4e5f\u662f\u6211\u4eec\u5e94\u8be5\u52aa\u529b\u7684\u65b9\u5411\uff0c\u8fd9\u5c31\u662f RL \u7684\u610f\u4e49\u6240\u5728\uff0c\u800c\u8fd9\u4e00\u70b9\u6b63\u662f LLM \u6240\u5b8c\u5168\u7f3a\u4e4f\u7684\u3002<\/p>\n<p style=\"text-align:center\" class=\"heading-2 ace-line old-record-id-QmzmdT9XNofYqSxXtFfc7j2lnVe\"><span style=\"font-size: 20px\"><strong>02 \u5f3a\u5316\u5b66\u4e60\u7684\u7a81\u7834\u8def\u5f84<\/strong><\/span><\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u4f60\u5bf9\u7ecf\u9a8c\uff08experience\uff09\u7684\u5b9a\u4e49\u662f\u4ec0\u4e48\uff1f\u6211\u8ba4\u4e3a\u56fe\u7075\u5728\u64b0\u5199\u90a3\u7bc7\u8bba\u6587\u65f6\u6709\u7740\u4ed6\u81ea\u5df1\u5bf9\u7ecf\u9a8c\u7684\u7406\u89e3\uff1b\u53e6\u5916\uff0c\u50cf AlphaGo \u90a3\u6837\uff0c\u901a\u8fc7\u81ea\u6211\u5bf9\u5f08\u751f\u6210\u4e00\u7cfb\u5217\u8f68\u8ff9\u6570\u636e\uff0c\u8fd9\u4e5f\u7b97\u4e00\u79cd\u7ecf\u9a8c\uff0c\u800c\u73b0\u5728\u60c5\u51b5\u53d1\u751f\u4e86\u4ec0\u4e48\u53d8\u5316\u5462\uff1f\u4f60\u8ba4\u4e3a\u5982\u4eca\u7684\u7ecf\u9a8c\u548c\u8fc7\u53bb\u4e00\u6837\u8fd8\u662f\u6709\u6240\u4e0d\u540c\u4e86\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a \u6211\u8ba4\u4e3a\u7ecf\u9a8c\u8fd9\u4e00\u6982\u5ff5\u975e\u5e38\u81ea\u7136\uff0c\u5e76\u4e14\u968f\u7740\u65f6\u95f4\u63a8\u79fb\u4e00\u76f4\u4ee5\u6765\u90fd\u662f\u5982\u6b64\u3002\u4f60\u7684\u7ecf\u9a8c\u5e76\u975e\u67d0\u79cd\u602a\u5f02\u3001\u6a21\u7cca\u6216\u96be\u4ee5\u6349\u6478\u7684\u4e1c\u897f\uff0c\u5b83\u5c31\u662f\u4f60<strong>\u5728\u4e16\u754c\u4e2d\u751f\u5b58\u65f6\u63a5\u6536\u5230\u7684\u771f\u5b9e\u6570\u636e<\/strong>\u3002\u4f60\u91c7\u53d6\u884c\u52a8\u3001\u505a\u5404\u79cd\u4e8b\u60c5\uff0c\u7136\u540e\u770b\u5230\u7ed3\u679c\u2014\u2014\u8fd9\u4e9b\u5c31\u662f\u4f60\u7684\u89c2\u5bdf\uff0c\u540c\u65f6\u4f60\u4e5f\u4f1a\u611f\u53d7\u5230\u5956\u52b1\u3002\u8fd9\u4e09\u7ec4\u65f6\u95f4\u5e8f\u5217\u5171\u540c\u6784\u6210\u4e86\u4f60\u7684\u7ecf\u9a8c\u3002<\/p>\n<p>\u5b9e\u9645\u4e0a\uff0c<strong>\u81f3\u5c11\u4ece agent \u7684\u89c6\u89d2\u4ee5\u53ca\u5728 RL \u7684\u601d\u7ef4\u6846\u67b6\u4e0b\uff0c\u9664\u4e86\u4f60\u63a5\u6536\u548c\u751f\u6210\u7684\u6570\u636e\u4e4b\u5916\uff0c\u518d\u65e0\u5176\u4ed6\u4e1c\u897f<\/strong>\u3002\u800c\u4e14\uff0c\u4f60\u65e0\u6cd5\u5148\u9a8c\u5730\u53bb\u8c08\u8bba\u7a7a\u95f4\u3001\u7269\u4f53\u3001\u7269\u7406\u89c4\u5f8b\u6216\u51e0\u4f55\u6982\u5ff5\uff0c\u56e0\u4e3a\u4f60\u65e0\u6cd5\u76f4\u63a5\u63a5\u89e6\u5230\u8fd9\u4e9b\u4e1c\u897f\uff1b\u4f60\u80fd\u76f4\u63a5\u63a5\u89e6\u7684\uff0c\u53ea\u6709\u4f60\u7684\u89c2\u5bdf\u548c\u884c\u52a8\u3002\u800c\u90a3\u4e9b\u6982\u5ff5\uff0c\u4e0d\u8fc7\u662f\u4f60\u4e3a\u4e86\u89e3\u91ca\u81ea\u8eab\u7ecf\u9a8c\u3001\u4e3a\u7ecf\u9a8c\u5efa\u7acb\u6a21\u578b\u800c\u7f16\u9020\u7684\u6545\u4e8b\u7f62\u4e86\u3002<\/p>\n<p>\u7ecf\u9a8c\u662f\u6839\u672c\u6027\u7684\uff0c\u5b83\u662f\u667a\u80fd\u7684\u6838\u5fc3\u3002\u6211\u4eec\u6709\u65f6\u4f1a\u8bf4\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\uff0c\u4f46\u540c\u6837\u91cd\u8981\u7684\u662f\uff0c\u4f60\u7684\u77e5\u8bc6\u672c\u8eab\u5c31\u4e0e\u7ecf\u9a8c\u76f8\u5173\u3002\u6bd4\u5982\uff0c\u5f53\u4f60\u76f8\u4fe1\u67d0\u4ef6\u4e8b\u65f6\uff0c\u8fd9\u4e2a\u4fe1\u5ff5\u672c\u8d28\u4e0a\u662f\u5173\u4e8e\u672a\u6765\u4f60\u4f1a\u63a5\u6536\u5230\u54ea\u4e9b\u89c2\u5bdf\u7684\u9884\u5224\u3002\u8bf4\u5230\u5e95\uff0c\u5b83\u5c31\u662f\u4e00\u79cd\u7406\u8bba\uff1a\u5173\u4e8e\u4f60\u672a\u6765\u4f1a\u6536\u5230\u4ec0\u4e48\u6570\u636e\u3001\u4f1a\u62e5\u6709\u600e\u6837\u7684\u7ecf\u9a8c\u7684\u7406\u8bba\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u4f60\u89c9\u5f97\u6837\u672c\u590d\u6742\u5ea6\u662f\u4eba\u5de5\u667a\u80fd\u9886\u57df\u91cc\u6211\u4eec\u9700\u8981\u89e3\u51b3\u7684\u4e00\u4e2a\u95ee\u9898\u5417\uff1f\u8fd8\u662f\u8bf4\uff0c\u4f60\u5176\u5b9e\u5e76\u4e0d\u8ba4\u4e3a\u8fd9\u662f\u4e2a\u95ee\u9898\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u6211\u4eec\u7684\u7b97\u6cd5\u5b66\u4e60\u6548\u7387\u786e\u5b9e\u4e0d\u591f\u9ad8\uff0c\u4f46\u6211\u5e76\u4e0d\u89c9\u5f97\u8fd9\u662f RL \u72ec\u6709\u7684\u95ee\u9898\u3002\u5b9e\u9645\u4e0a\uff0c\u6211\u4eec\u6240\u6709\u7684\u7b97\u6cd5\u5b66\u4e60\u6548\u7387\u90fd\u5f88\u4f4e\uff0c\u6240\u4ee5\u5fc5\u987b\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\u3002\u8bf4\u5230\u5e95\uff0c\u6216\u8bb8\u6574\u4e2a\u7814\u7a76\u7684\u6838\u5fc3\u5c31\u662f\u8981\u4ece\u7ecf\u9a8c\u4e2d\u9ad8\u6548\u5b66\u4e60\u2014\u2014\u8fd9\u6b63\u662f\u6211\u4eec\u7814\u7a76\u8005\u6b63\u5728\u505a\u7684\u4e8b\uff0c\u6211\u4eec\u5728\u52aa\u529b\u5bfb\u627e\u80fd\u8ba9\u6211\u4eec\u4ece\u7ecf\u9a8c\u4e2d\u5c3d\u53ef\u80fd\u9ad8\u6548\u5b66\u4e60\u7684\u7b97\u6cd5\u3002<\/p>\n<p>\u5728\u67d0\u79cd\u610f\u4e49\u4e0a\uff0c\u4ece\u5956\u52b1\u4e2d\u5b66\u4e60\u6bd4\u4ece\u6307\u4ee4\u4e2d\u5b66\u4e60\u66f4\u56f0\u96be\uff0c\u5982\u679c\u4e16\u754c\u76f4\u63a5\u544a\u8bc9\u4f60\u8be5\u505a\u4ec0\u4e48\uff0c\u5b66\u4e60\u8d77\u6765\u4f1a\u7b80\u5355\u5f97\u591a\u3002\u4f46\u663e\u7136\uff0c<strong>\u4e16\u754c\u4e0d\u4f1a\u76f4\u63a5\u7ed9\u51fa\u6307\u4ee4\uff0c\u4e16\u754c\u53ea\u4f1a\u7ed9\u4f60\u629b\u51fa\u4e00\u4e2a\u95ee\u9898<\/strong>\uff0c\u4f60\u5f97\u81ea\u5df1\u5f04\u660e\u767d\u8be5\u600e\u4e48\u505a\uff0c\u6240\u4ee5\u8fd9\u624d\u662f\u771f\u6b63\u7684\u95ee\u9898\u6240\u5728\u3002\u56e0\u6b64\uff0c\u6211\u4e0d\u4f1a\u628a\u8fd9\u79cd\u95ee\u9898\u770b\u4f5c\u662f\u4e00\u79cd\u56f0\u96be\uff0c\u800c\u662f\u4e00\u9879\u6311\u6218\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u8fd9\u4e24\u4e2a\u5355\u4e00\u7684\u5956\u52b1\u4f1a\u662f\u8fd9\u4e2a\u6807\u91cf\u5417\uff1f\u8981\u77e5\u9053\uff0c\u5956\u52b1\u672c\u8eab\u4e5f\u5206\u4e0d\u540c\u79cd\u7c7b\u7684<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u8fd9\u5c31\u662f\u6211\u4eec\u6240\u8bf4\u7684\u201c\u5956\u52b1\u5047\u8bbe\u201d\u4e86\u3002<strong>\u6211\u4eec\u6240\u8bf4\u7684\u76ee\u6807\u4e0e\u610f\u56fe\uff0c\u90fd\u53ef\u4ee5\u88ab\u5f88\u597d\u5730\u7406\u89e3\u4e3a\uff1a\u5bf9\u4e00\u79cd\u540d\u4e3a \u201c\u5956\u52b1\u201d \u7684\u6807\u91cf\u4fe1\u53f7\u7684\u7d2f\u79ef\u548c\u7684\u671f\u671b\u503c\u8fdb\u884c\u6700\u5927\u5316\uff0c\u8fd9\u662f\u5bf9\u667a\u80fd\u95ee\u9898\u4e00\u9879\u6781\u5177\u4ef7\u503c\u7684\u7b80\u5316\u4e0e\u63d0\u70bc<\/strong>\u3002\u8981\u77e5\u9053\uff0c\u667a\u80fd\u5c31\u662f\u4e0e\u4e16\u754c\u4ea4\u4e92\u4ee5\u8fbe\u6210\u76ee\u6807\uff0c\u800c\u4f60\u9700\u8981\u67d0\u79cd\u65b9\u5f0f\u6765\u5b9a\u4e49\u8fd9\u4e2a\u76ee\u6807\u3002<\/p>\n<p>\u5728\u5f3a\u5316\u5b66\u4e60\u4e2d\uff0c\u6211\u4eec\u5efa\u8bae\uff1a<strong>\u6240\u6709\u76ee\u6807\u90fd\u53ef\u4ee5\u88ab\u89e3\u8bfb\u4e3a \u201c\u5956\u52b1\u201d\uff0c\u4e00\u79cd\u5355\u4e00\u7684\u6807\u91cf\u6570\u503c<\/strong>\u3002<\/p>\n<p>\u5728\u4f60\u4eec\u770b\u6765\uff0c\u8fd9\u6837\u8ba4\u4e3a\u53ef\u80fd\u663e\u5f97\u6709\u4e9b\u6d45\u8584\uff0c\u751a\u81f3\u683c\u5c40\u4e0d\u5927\u3002\u4f46\u6211\u8ba4\u4e3a\uff0c\u8fd9\u6070\u6070\u662f\u5b83\u6781\u5177\u667a\u6167\u7684\u90e8\u5206\u2014\u2014\u5b83\u975e\u5e38\u6e05\u6670\uff0c\u9ad8\u5ea6\u7b80\u5316\uff0c\u5374\u53c8\u4f3c\u4e4e\u8db3\u591f\u5b8c\u5907\u3002\u4e8b\u5b9e\u4e0a\uff0c\u50cf\u98ce\u9669\u8003\u91cf\u3001\u591a\u76ee\u6807\u8fd9\u7c7b\u66f4\u590d\u6742\u7684\u8bbe\u5b9a\uff0c\u672a\u5fc5\u80fd\u8ba9\u7cfb\u7edf\u53d8\u5f97\u66f4\u5f3a\u5927\u3002\u963f\u5c14\u4f2f\u5854\u5927\u5b66\uff08\u4e5f\u5c31\u662f\u6211\u6240\u5728\u7684\u5927\u5b66\uff09\u7684\u51e0\u4f4d\u7814\u7a76\u8005\u53d1\u8868\u8fc7\u4e00\u7bc7\u5f88\u68d2\u7684\u8bba\u6587\uff0c\u540d\u4e3a\u300aSettling the Reward Hypothesis\u300b\uff0c\u8fd9\u7bc7\u8bba\u6587\u6df1\u5165\u63a2\u8ba8\u4e86\u8fd9\u4e00\u5047\u8bf4\uff0c\u8bba\u8bc1\u4e86\u6211\u4eec\u5e76\u4e0d\u9700\u8981\u591a\u76ee\u6807\u3001\u98ce\u9669\u7b49\u590d\u6742\u8bbe\u5b9a\uff0c\u5355\u4e00\u5956\u52b1\u4fe1\u53f7\u5df2\u8db3\u591f\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-6d838055d4078cf6c79e19d40ae1c90d.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-6d838055d4078cf6c79e19d40ae1c90d.png\"><\/a><\/p>\n<p><em>\u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/pdf\/2212.10420<\/em><\/p>\n<p>\u6211\u8fd8\u8bb0\u5f97\u81ea\u5df1\u6700\u521d\u771f\u6b63\u63a5\u53d7\u8fd9\u4e00\u5047\u8bbe\u65f6\u7684\u60c5\u666f\uff0c\u5b83\u786e\u5b9e\u663e\u5f97\u6709\u4e9b\u683c\u5c40\u4e0d\u5927\u3002\u6bd5\u7adf\uff0c\u6211\u4eec\u7684\u4eba\u751f\u4f3c\u4e4e\u627f\u8f7d\u7740\u66f4\u5b8f\u5927\u7684\u76ee\u6807\uff1a\u6bd4\u5982\u62ef\u6551\u4e16\u754c\u3001\u79ef\u7d2f\u8d22\u5bcc\u3001\u4fdd\u62a4\u751f\u6001\uff0c\u6216\u8005\u517b\u80b2\u5bb6\u5ead\u3002\u6211\u4eec\u6709\u8bb8\u591a\u76ee\u6807\uff0c\u800c\u4e14\u5b83\u4eec\u4f3c\u4e4e\u90fd\u4e0e\u73b0\u5b9e\u4e16\u754c\u7d27\u5bc6\u76f8\u5173\u3001\u5177\u4f53\u53ef\u611f\u3002\u4f46\u8981\u6253\u9020\u7b97\u6cd5\u3001\u6784\u5efa\u667a\u80fd\uff0c\u5c31\u5fc5\u987b\u5c06\u95ee\u9898\u8f6c\u5316\u4e3a\u4e00\u79cd\u6807\u51c6\u5f62\u5f0f\u3002\u5956\u52b1\u5047\u8bbe\u6b63\u662f\u8fd9\u6837\u4e00\u4e2a\u6781\u5177\u4ef7\u503c\u7684\u6838\u5fc3\u601d\u60f3\uff0c\u5b83\u975e\u5e38\u6df1\u523b\uff0c\u6781\u5927\u5730\u5e2e\u52a9\u6211\u4eec\u7406\u89e3\u4e86\u667a\u80fd\u7a76\u7adf\u662f\u4ec0\u4e48\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u8fd9\u79cd\u6574\u5408\u65b9\u5f0f\u80fd\u5426\u89e3\u51b3\u201c\u5956\u52b1\u96be\u4ee5\u5b9a\u4e49\u201d\u6216\u201c\u5956\u52b1\u5b9a\u4e49\u5f97\u975e\u5e38\u6a21\u7cca\u201d\u7684\u95ee\u9898\u5462\uff1f\u4f60\u8ba4\u4e3a\uff0c\u660e\u786e\u5956\u52b1\u662f\u5426\u5df2\u6210\u4e3a\u4eba\u5de5\u667a\u80fd\u6216\u4efb\u4f55\u5b66\u4e60\u7b97\u6cd5\u7684\u4e00\u79cd\u5148\u51b3\u6761\u4ef6\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u9996\u5148\uff0c\u6211\u4eec\u5148\u56de\u5f52\u81ea\u7136\u3001\u52a8\u7269\u548c\u4eba\u7c7b\u672c\u8eab\u3002\u6ca1\u4eba\u9700\u8981\u4e3a\u6211\u4eec\u5236\u5b9a\u5956\u52b1\uff0c\u6211\u4eec\u7684\u5956\u52b1\u662f\u8fdb\u5316\u7684\u4ea7\u7269\u3002\u90a3\u6211\u4eec\u7684\u5956\u52b1\u5230\u5e95\u662f\u4ec0\u4e48\uff1f\u5e76\u6ca1\u6709\u4e00\u6839\u7ebf\u4ece\u5916\u90e8\u63a5\u5165\u6211\u4eec\u7684\u5927\u8111\uff0c\u5956\u52b1\u7a76\u7adf\u5b58\u5728\u4e8e\u4f55\u5904\u5462\uff1f\u5b9e\u9645\u4e0a\uff0c<strong>\u5956\u52b1\u662f\u5728\u4f60\u7684\u9885\u9aa8\u5185\u90e8\u8ba1\u7b97\u4ea7\u751f\u7684<\/strong>\u3002\u5927\u8111\u4e2d\u6709\u4e00\u4e2a\u533a\u57df\uff0c\u6211\u8ba4\u4e3a\u662f\u4e0b\u4e18\u8111\uff0c\u5b83\u4f1a\u8ba1\u7b97\u4e00\u7cfb\u5217\u72b6\u6001\uff1a\u4f60\u7684\u8eab\u4f53\u72b6\u51b5\u662f\u5426\u826f\u597d\uff1f\u8840\u6db2\u4e2d\u662f\u5426\u542b\u6709\u8db3\u591f\u7684\u8425\u517b\uff1f\u4f60\u662f\u5426\u8131\u79bb\u4e86\u75db\u82e6\uff1f\u662f\u5426\u611f\u5230\u4e0d\u9002\uff1f\u4f53\u6e29\u662f\u5426\u9002\u5b9c\uff1f\u6240\u6709\u8fd9\u4e9b\u56e0\u7d20\u90fd\u4f1a\u5f71\u54cd\u4f60\u7684\u5956\u52b1\u4fe1\u53f7\u3002<\/p>\n<p>\u6240\u4ee5\uff0c\u5956\u52b1\u4fe1\u53f7\u5b58\u5728\u4e8e\u4f60\u7684\u9885\u9aa8\u4e4b\u5185\uff0c\u5b83\u662f\u4f60\u5927\u8111\uff08\u5b9e\u9645\u4e0a\u4e5f\u662f\u4f60\u8eab\u4f53\uff09\u7684\u4e00\u90e8\u5206\u3002\u4f46\u6211\u4eec\u4e0d\u4f1a\u628a\u8eab\u4f53\u89c6\u4e3a\u73af\u5883\u7684\u4e00\u90e8\u5206\uff0c\u800c\u5927\u8111\u4e2d\u8fd9\u4e2a\u8d1f\u8d23\u8ba1\u7b97\u5956\u52b1\u7684\u5c0f\u533a\u57df\uff0c\u5374\u5c5e\u4e8e\u73af\u5883\u7684\u8303\u7574\u3002\u6211\u4eec\u5fc5\u987b\u5c06\u5176\u5f52\u4e3a\u73af\u5883\u7684\u4e00\u90e8\u5206\uff0c\u8fd9\u6837\u5b83\u624d\u4e0d\u4f1a\u53d7\u5230 agent \u7684\u76f4\u63a5\u63a7\u5236\uff0cagent \u65e0\u6cd5\u968f\u5fc3\u6240\u6b32\u5730\u5c06\u81ea\u8eab\u5956\u52b1\u8c03\u81f3\u9ad8\u4f4d\uff0c\u8981\u8ba9\u5956\u52b1\u5347\u9ad8\uff0c\u5b83\u5fc5\u987b\u91c7\u53d6\u884c\u52a8\uff1a\u6bd4\u5982\u907f\u514d\u75db\u82e6\u3001\u83b7\u53d6\u98df\u7269\u4ee5\u7ef4\u6301\u8840\u7cd6\u5728\u6ee1\u610f\u6c34\u5e73\uff0c\u5b83\u5fc5\u987b\u901a\u8fc7\u884c\u52a8\u5f71\u54cd\u5468\u56f4\u7684\u4e16\u754c\uff0c\u8fd9\u4e9b\u5f71\u54cd\u518d\u901a\u8fc7\u8eab\u4f53\u53cd\u9988\uff0c\u6700\u7ec8\u4ea7\u751f\u9ad8\u5956\u52b1\u4fe1\u53f7\u3002<\/p>\n<p>\u5728\u90a3\u4e4b\u540e\uff0c\u5f53\u6211\u4eec\u6253\u9020\u5de5\u7a0b\u5316\u7cfb\u7edf\u3001\u5e0c\u671b\u4eba\u5de5\u667a\u80fd\u5b8c\u6210\u7279\u5b9a\u4efb\u52a1\u65f6\uff0c\u4f5c\u4e3a\u8bbe\u8ba1\u8005\uff0c\u6211\u4eec\u5fc5\u987b\u60f3\u6e05\u695a\u3001\u660e\u786e\u81ea\u5df1\u771f\u6b63\u5e0c\u671b\u8fbe\u6210\u7684\u76ee\u6807\u662f\u4ec0\u4e48\uff0c\u5e76\u4e14\u5f97\u627e\u5230\u65b9\u6cd5\uff0c\u5c06\u8fd9\u4e2a\u76ee\u6807\u8f6c\u5316\u4e3a\u5956\u52b1\u4fe1\u53f7\u4f20\u9012\u7ed9 agent\u3002\u8fd9\u4e00\u6b65\u6709\u65f6\u9887\u5177\u6311\u6218\uff0c\u6709\u65f6\u53c8\u6bd4\u8f83\u7b80\u5355\u3002\u6bd4\u5982\uff0c\u4e0b\u56fd\u9645\u8c61\u68cb\uff0c\u8d62\u5f97\u6bd4\u8d5b\u4f60\u4f1a\u5f97\u5230\u5956\u52b1\uff1b\u4e0b\u56f4\u68cb\uff0c\u8d62\u5f97\u6bd4\u8d5b\u4e5f\u4f1a\u5f97\u5230\u5956\u52b1\u3002\u5982\u679c\u662f\u4f01\u4e1a\u60f3\u8981\u76c8\u5229\uff0c\u4f60\u53ef\u4ee5\u4e3a\u6210\u529f\u8d5a\u94b1\u7684\u4eba\u63d0\u4f9b\u5956\u52b1\u3002\u80fd\u4f5c\u4e3a\u5956\u52b1\u7684\u4e1c\u897f\u6709\u5f88\u591a\uff0c\u6709\u65f6\u5f88\u7b80\u5355\uff0c\u6709\u65f6\u5374\u771f\u7684\u5f88\u96be\u3002\u96be\u70b9\u90e8\u5206\u5728\u4e8e\uff0c<strong>\u6211\u4eec\u81ea\u5df1\u5f80\u5f80\u90fd\u96be\u4ee5\u786e\u5b9a\u6700\u60f3\u8ba9\u4eba\u5de5\u667a\u80fd\u5b8c\u6210\u7684\u5230\u5e95\u662f\u4ec0\u4e48<\/strong>\u3002<\/p>\n<p>\u6211\u60f3\u8bf4\u7684\u662f\uff1a\u4eba\u4eec\u5e38\u5e38\u6709\u4e00\u79cd\u5f3a\u70c8\u7684\u503e\u5411\uff0c\u5c31\u662f\u628a\u95ee\u9898\u7684\u89e3\u51b3\u65b9\u6848\u76f4\u63a5\u585e\u8fdb\u5956\u52b1\u91cc\u3002\u6bd4\u5982\uff0c\u60f3\u8ba9 AI \u4e0b\u56f4\u68cb\uff0c\u4f60\u53ef\u80fd\u4f1a\u5fcd\u4e0d\u4f4f\u5728\u8fc7\u7a0b\u4e2d\u8bbe\u7f6e\u4e00\u4e9b\u989d\u5916\u5956\u52b1\uff0c\u6bd4\u5982\u63a7\u5236\u68cb\u76d8\u7684\u7279\u5b9a\u533a\u57df\u6216\u505a\u51fa\u773c\u4f4d\u5c31\u7ed9\u5956\u52b1\uff0c\u4f46\u8fd9\u51e0\u4e4e\u662f\u9519\u8bef\u7684\u3002<strong>\u4f60\u5e94\u8be5\u4e3a\u81ea\u5df1\u771f\u6b63\u60f3\u8981\u7684\u7ed3\u679c\u8bbe\u7f6e\u5956\u52b1<\/strong>\uff0c\u5bf9\u4e8e\u56f4\u68cb\u800c\u8a00\uff0c\u771f\u6b63\u60f3\u8981\u7684\u5c31\u662f\u8d62\u68cb\uff0c\u505a\u51fa\u773c\u4f4d\u53ea\u662f\u901a\u5411\u8d62\u68cb\u7684\u4e00\u6761\u8def\u5f84\u3002\u5c3d\u7ba1\u4f60\u53ef\u80fd\u662f\u60f3\u5e2e\u4eba\u5de5\u667a\u80fd\u4e00\u628a\uff0c\u8fd9\u79cd\u5c1d\u8bd5\u4e5f\u975e\u5e38\u666e\u904d\uff0c\u4f46\u8fd9\u6837\u5b9e\u9645\u4e0a\u6539\u53d8\u4e86\u95ee\u9898\u672c\u8eab\uff0c\u800c\u6211\u4eec\u5e76\u4e0d\u60f3\u6539\u53d8\u95ee\u9898\u3002\u5982\u679c\u4f60\u60f3\u8d62\uff0c\u5c31\u5e94\u8be5\u53ea\u4e3a\u8d62\u8bbe\u7f6e\u5956\u52b1\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u4f60\u4e4b\u524d\u63d0\u5230\u8fc7\u8981\u4fdd\u6301\u5f00\u653e\u7684\u5fc3\u6001\uff0c\u5e76\u4e14\u4e5f\u9f13\u52b1\u90a3\u4e9b\u521a\u62ff\u5230\u535a\u58eb\u5b66\u4f4d\u7684\u5e74\u8f7b\u7814\u7a76\u8005\u8fd9\u6837\u505a\uff0c\u6211\u6700\u8fd1\u770b\u4e86\u4e00\u4e9b\u7814\u7a76\u63d0\u6848\uff0c\u5176\u4e2d\u5173\u4e8e\u6301\u7eed\u5b66\u4e60\u7684\u4e00\u4e9b\u60f3\u6cd5\u6211\u5f88\u559c\u6b22\uff0c\u6240\u4ee5\u4f60\u80fd\u4e0d\u80fd\u8fdb\u4e00\u6b65\u8bb2\u8bb2\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u65e0\u8bba\u662f\u963f\u5c14\u4f2f\u5854\u7701\u8fd8\u662f\u65b0\u52a0\u5761\u7684\u7814\u7a76\u9879\u76ee\u90fd\u662f\u4ece RL \u5f00\u59cb\u7684\uff0c\u5c24\u5176\u662f\u201c\u963f\u5c14\u4f2f\u5854\u4eba\u5de5\u667a\u80fd\u7814\u7a76\u8ba1\u5212\u201d\uff08The Alberta Plan for AI Research\uff09\uff0c\u8fd9\u662f\u4e00\u4efd\u6587\u4ef6\uff0c\u4f60\u53ef\u4ee5\u5728 arXiv \u4e0a\u627e\u5230\u5b83\u3002\u8fd9\u4efd\u6587\u4ef6\u7531 Michael Bowling\u3001Patrick Plarskin \u548c\u6211\u81ea\u5df1\u5171\u540c\u64b0\u5199\u3002\u5728\u5176\u4e2d\uff0c\u6211\u4eec\u8bd5\u56fe\u5c3d\u53ef\u80fd\u6e05\u6670\u5730\u52fe\u52d2\u51fa\u4ece\u5f53\u524d\u9636\u6bb5\u8fc8\u5411 full intelligent agent\uff0c\u5373 the full AI \u6240\u9700\u7684\u6b65\u9aa4\u3002\u8fd9\u662f\u4e00\u4e2a\u5305\u542b 12 \u4e2a\u6b65\u9aa4\u7684\u8ba1\u5212\uff0c\u8fd9\u4e9b\u6b65\u9aa4\u5e76\u975e\u5fc5\u987b\u6309\u987a\u5e8f\u6267\u884c\uff0c\u4f46\u6211\u76ee\u524d\u6b63\u81f4\u529b\u4e8e\u7b2c\u4e00\u6b65\uff0c\u5e76\u4e14\u70ed\u5207\u671f\u5f85\u5b8c\u6210\u7b2c\u4e00\u6b65\u540e\u80fd\u5f00\u59cb\u7740\u624b\u7b2c\u4e8c\u6b65\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-aea190c71a8e2d1657cabf99b4672205.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-aea190c71a8e2d1657cabf99b4672205.png\"><\/a><\/p>\n<p><em>\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/pdf\/2208.11173<\/em><\/p>\n<p>\u8ba1\u5212\u7684\u524d\u671f\u6b65\u9aa4\u90fd\u4e0e\u4f7f\u7528\u51fd\u6570\u8fd1\u4f3c\uff08function approximation\uff09\u8fdb\u884c\u9ad8\u6548\u5b66\u4e60\u76f8\u5173\uff0c\u5b9e\u9645\u4e0a\u4e5f\u90fd\u56f4\u7ed5 RL \u5c55\u5f00\uff0c\u6d89\u53ca\u6301\u7eed\u5b66\u4e60\uff08continual learning\uff09\u3001\u5143\u5b66\u4e60\uff08meta-learning\uff09\u7b49\u65b9\u5411\u2014\u2014\u771f\u6b63\u805a\u7126 RL \u672c\u8eab\u7684\u5185\u5bb9\u5176\u5b9e\u4ece\u7b2c\u4e09\u6b65\u624d\u5f00\u59cb\u3002\u4f46\u5f52\u6839\u7ed3\u5e95\uff0c<strong>\u6574\u4e2a\u8ba1\u5212\u7684\u76ee\u6807\u662f\u6253\u9020\u4e00\u4e2a RL agent\uff1a\u4e00\u4e2a\u5177\u5907\u63a8\u7406\u3001\u89c4\u5212\u4e0e\u5b66\u4e60\u80fd\u529b\u7684\u5b8c\u6574 agent\u3002<\/strong><\/p>\n<p>\u6b64\u5916\uff0c\u8be5\u8ba1\u5212\u5b8c\u5168\u4ee5\u5956\u52b1\u4e3a\u6838\u5fc3\u57fa\u7840\uff0c\u4f46\u5728\u5b9e\u8df5\u8fc7\u7a0b\u4e2d\uff0c\u4f1a\u884d\u751f\u51fa\u5b50\u76ee\u6807\u6216\u8f85\u52a9\u4efb\u52a1\uff0c\u8fd9\u4e9b\u5b50\u76ee\u6807\u53ef\u4ee5\u662f\u6211\u4eec\u65e5\u5e38\u4f1a\u8003\u8651\u7684\u4e8b\u60c5\uff0c\u6bd4\u5982\u62ff\u8d77\u4e00\u4e2a\u7269\u4f53\u53ef\u80fd\u6210\u4e3a\u4e00\u4e2a\u5b50\u4efb\u52a1\uff0c\u9ad8\u6548\u884c\u8d70\u662f\u4e00\u4e2a\u76ee\u6807\uff0c\u4e0b\u56f4\u68cb\u4e5f\u53ef\u4ee5\u662f\u4e00\u4e2a\u9700\u8981\u8fbe\u6210\u7684\u5b50\u76ee\u6807\uff0c\u751a\u81f3\u83b7\u5f97\u5b66\u4f4d\u3002\u57fa\u672c\u4e0a\u4f60\u4e00\u6574\u5929\u505a\u7684\u6240\u6709\u4e8b\uff0c\u6b65\u884c\u53bb\u5927\u5b66\u3001\u559d\u676f\u6c34\u3001\u627e\u6d17\u624b\u95f4\uff0c\u8fd9\u4e9b\u90fd\u53ef\u4ee5\u6210\u4e3a\u4f60\u52aa\u529b\u8fbe\u6210\u5e76\u5b66\u4e60\u76f8\u5173\u6280\u80fd\u7684\u76ee\u6807\u3002\u800c\u6240\u6709\u8fd9\u4e9b\u5b50\u76ee\u6807\u548c\u6280\u80fd\uff0c\u6700\u7ec8\u90fd\u670d\u52a1\u4e8e\u90a3\u4e00\u4e2a\u6838\u5fc3\u76ee\u6807\uff0c\u4f8b\u5982\u5b8c\u6210\u5de5\u4f5c\uff0c\u8fd9\u5c31\u662f\u963f\u5c14\u4f2f\u5854\u4eba\u5de5\u667a\u80fd\u7814\u7a76\u8ba1\u5212\u7684\u4e00\u90e8\u5206\u5185\u5bb9\u3002<\/p>\n<p style=\"text-align:center\" class=\"heading-2 ace-line old-record-id-EbpXdfK1foCTNexMCoecb6Xqnwb\"><span style=\"font-size: 20px\"><strong>03 \u901a\u7528\u667a\u80fd\u4f53\u7684\u8ba4\u77e5\u8f6c\u5411<\/strong><\/span><\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u4f60\u63d0\u5230\u4e86\u5143\u5b66\u4e60\uff0c\u5173\u4e8e\u5982\u4f55\u8ba9 RL \u5728\u8de8\u4efb\u52a1\u573a\u666f\u4e0b\u66f4\u5177\u901a\u7528\u6027\uff0c\u4f60\u7684\u770b\u6cd5\u662f\u4ec0\u4e48\uff1f\u5982\u4f55\u8ba9\u5b83\u4e0d\u4ec5\u80fd\u89e3\u51b3\u96c5\u8fbe\u5229\u56f4\u68cb\u6e38\u620f\u8fd9\u7c7b\u7279\u5b9a\u4efb\u52a1\uff0c\u8fd8\u80fd\u5177\u5907\u66f4\u5e7f\u6cdb\u7684\u901a\u7528\u6027\uff1f\u4f60\u8ba4\u4e3a\u5143\u5b66\u4e60\u4f1a\u662f\u6253\u9020\u901a\u7528 agent \u7684\u6709\u529b\u5019\u9009\u65b9\u5411\u5417\uff1f\u8fd8\u662f\u8bf4\uff0c\u6211\u4eec\u9700\u8981\u5176\u4ed6\u5b8c\u5168\u4e0d\u540c\u7684\u65b9\u6cd5\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u6211\u4eec\u5f53\u7136\u60f3\u8981\u6253\u9020\u901a\u7528 agent\uff0c\u4f46\u5e76\u4e0d\u8ba4\u540c\u201ctask\u201d\uff08\u4efb\u52a1\uff09\u8fd9\u4e2a\u6982\u5ff5\u3002\u8bf4\u5b9e\u8bdd\uff0c\u6211\u89c9\u5f97\u6211\u4eec\u53ea\u662f\u751f\u6d3b\u800c\u5df2\uff0c\u6211\u4e0d\u89c9\u5f97\u6211\u6709\u4ec0\u4e48\u4efb\u52a1\u662f\u9700\u8981\u5355\u72ec\u5956\u52b1\u7684\uff0c\u6211\u6709\u81ea\u5df1\u7684\u4eba\u751f\uff0c\u6709\u81ea\u5df1\u7684\u5956\u52b1\u4fe1\u53f7\uff0c\u5b83\u4f1a\u4e00\u76f4\u6301\u7eed\u4e0b\u53bb\uff0c\u867d\u7136\u5e76\u975e\u6c38\u6052\uff0c\u4f46\u4f1a\u5ef6\u7eed\u5f88\u957f\u65f6\u95f4\u3002<\/p>\n<p>\u6bcf\u6b21\u542c\u5230\u4eba\u4eec\u8c08\u8bba\u201ctask\u201d\uff0c\u6211\u5176\u5b9e\u90fd\u4e0d\u592a\u6e05\u695a\u4ed6\u4eec\u5177\u4f53\u6307\u4ec0\u4e48\uff0c\u4f46\u6211\u5f88\u6e05\u695a\u201cskill\u201d\uff08\u6280\u80fd\uff09\u662f\u4ec0\u4e48\u3002\u6bd4\u5982\uff0c\u6211\u6709\u719f\u7ec3\u884c\u8d70\u7684\u6280\u80fd\uff0c\u6709\u627e\u5230\u53bb\u94f6\u884c\u8def\u7ebf\u7684\u6280\u80fd\uff0c\u8fd8\u6709\u505a\u5176\u4ed6\u5404\u79cd\u4e8b\u60c5\u7684\u6280\u80fd\u3002\u8fd9\u4e9b\u5c31\u50cf\u6211\u4e4b\u524d\u63d0\u5230\u7684\u8f85\u52a9\u6027\u95ee\u9898\uff1a\u62ff\u8d77\u4e00\u4e2a\u7269\u4f53\u3001\u627e\u5230\u53bb\u4e0d\u540c\u5730\u65b9\u7684\u8def\uff0c\u7b49\u7b49\u3002\u6211\u4f1a\u5b66\u4e60\u5982\u4f55\u5b8c\u6210\u6240\u6709\u8fd9\u4e9b\u4e8b\uff0c\u7136\u540e\u601d\u8003\u5982\u4f55\u5c06\u8fd9\u4e9b\u6280\u80fd\u7ec4\u5408\u8d77\u6765\uff0c\u4ee5\u83b7\u5f97\u5c5e\u4e8e\u81ea\u5df1\u7684\u5956\u52b1\u3002<\/p>\n<p>\u901a\u7528\u6027\u81f3\u5173\u91cd\u8981\uff0c\u6211\u4eec\u786e\u5b9e\u8ffd\u6c42\u901a\u7528\u6027\uff0c\u4f46\u6211\u4eec\u4e0d\u60f3\u5728\u201ctask\u201d\u4e4b\u95f4\u8ffd\u6c42\u901a\u7528\u3002\u56e0\u4e3a\u201ctask\u201d\u5e76\u975e\u8fd9\u4e00\u6846\u67b6\u7684\u7ec4\u6210\u90e8\u5206\uff0c\u4e5f\u4e0d\u5c5e\u4e8e RL \u7684\u8303\u7574\uff0c\u5e76\u4e0d\u5b58\u5728\u201ctask\u201d\u8fd9\u79cd\u6982\u5ff5\uff0c\u8fd9\u91cc\u53ea\u6709\u5355\u4e00\u7684\u5956\u52b1\u4fe1\u53f7\uff0c\u4ec5\u6b64\u800c\u5df2\u3002\u4f46\u6211\u4eec\u786e\u5b9e\u9700\u8981\u63a2\u8ba8\u6cdb\u5316\uff08generalization\uff09\u3002\u8fd9\u91cc\u7684\u6cdb\u5316\u6307\u7684\u662f<strong>\u72b6\u6001\u4e4b\u95f4\u7684\u6cdb\u5316\uff0c\u5728\u8fd9\u4e2a\u72b6\u6001\u4e0b\u5b66\u4f1a\u4e86\u8be5\u505a\u4ec0\u4e48\uff0c\u8fd9\u4f1a\u5f71\u54cd\u6211\u5bf9\u5728\u53e6\u4e00\u4e2a\u72b6\u6001\u4e0b\u8be5\u505a\u4ec0\u4e48\u624d\u6b63\u786e\u7684\u5224\u65ad\u3002<\/strong>\u6240\u4ee5\uff0c\u6cdb\u5316\u662f\u5728\u4e0d\u540c\u72b6\u6001\u95f4\u53d1\u751f\u7684\uff0c\u8fd9\u6b63\u662f\u6cdb\u5316\u7684\u6838\u5fc3\u4ef7\u503c\u6240\u5728\uff0c\u800c\u8fd9\u79cd\u6cdb\u5316\u80fd\u529b\u4e5f\u5c06\u52a9\u529b\u6211\u4eec\u6253\u9020\u51fa\u901a\u7528 agent\u3002<\/p>\n<p>\u6211\u4eec\u8bf4\u7684\u5143\u5b66\u4e60\u662f\u6307\uff1a\u5f53\u5b58\u5728\u4e00\u4e2a\u57fa\u7840\u5b66\u4e60\u8fc7\u7a0b\uff0c\u7136\u540e\u6211\u4eec\u5728\u8fd9\u4e2a\u57fa\u7840\u4e4b\u4e0a\u518d\u505a\u4e00\u4e9b\u64cd\u4f5c\u65f6\uff0c\u8c08\u8bba\u5143\u5b66\u4e60\u624d\u662f\u6070\u5f53\u7684\u3002\u6211\u8ba4\u4e3a\u6700\u4e3b\u8981\u7684\u4f8b\u5b50\u662f\u8fd9\u6837\u7684\uff1a\u6bd4\u5982\u6211\u4eec\u901a\u8fc7\u8c03\u6574\u6a21\u578b\u6743\u91cd\u6765\u89e3\u51b3\u67d0\u4e2a\u57fa\u7840\u5b66\u4e60\u4efb\u52a1\uff0c\u4e0d\u7ba1\u8fd9\u4e2a\u57fa\u7840\u4efb\u52a1\u662f\u76d1\u7763\u5b66\u4e60\uff08supervised learning\uff09\u8fd8\u662f RL\uff0c\u5728\u8fd9\u4e2a\u8fc7\u7a0b\u4e2d\uff0c\u6211\u4eec\u4f1a\u6d89\u53ca\u6b65\u957f\uff08step sizes\uff09\u4ee5\u53ca\u5176\u4ed6\u8d85\u53c2\u6570\u7684\u8bbe\u7f6e\u3002\u6240\u4ee5\uff0c\u8fd9\u4e9b\u6b65\u957f\u7684\u8c03\u6574\u5c31\u5c5e\u4e8e\u5143\u5b66\u4e60\u7684\u4e00\u4e2a\u4f8b\u5b50\u3002<\/p>\n<p>\u6211\u4eec\u7684\u6574\u4e2a\u7f51\u7edc\u90fd\u5e94\u8be5\u8bbe\u7f6e\u6b65\u957f\uff0c\u6bcf\u4e2a\u6743\u91cd\u90fd\u8be5\u6709\u81ea\u5df1\u4e13\u5c5e\u7684\u6b65\u957f\uff0c\u8fd9\u4e9b\u6b65\u957f\u5fc5\u987b\u901a\u8fc7\u81ea\u52a8\u5316\u6d41\u7a0b\u6765\u5355\u72ec\u8c03\u6574\uff0c\u800c\u4e0d\u662f\u7531\u4eba\u5de5\u64cd\u4f5c\u3002\u8fd9\u6837\u4e00\u6765\uff0c\u5982\u679c\u4f60\u80fd\u8c03\u6574\u8fd9\u4e9b\u6b65\u957f\uff0c\u5c31\u80fd\u4ece\u4e2d\u5f97\u77e5\uff1a\u8fd9\u4e9b\u6743\u91cd\u4e0d\u60f3\u8c03\u6574\uff0c\u90a3\u4e9b\u6743\u91cd\u60f3\u8c03\u6574\u3002\u800c\u5f53\u4f60\u51b3\u5b9a\u8c03\u6574\u54ea\u4e9b\u6743\u91cd\u65f6\uff0c\u8fd9\u4e9b\u6743\u91cd\u6b63\u662f\u4f60\u8981\u8fdb\u884c\u6cdb\u5316\u7684\u4f9d\u636e\u3002\u6253\u4e2a\u6bd4\u65b9\uff0c\u4f60\u6709\u5f88\u591a\u7279\u5f81\uff0c\u8be5\u57fa\u4e8e\u54ea\u4e9b\u7279\u5f81\u8fdb\u884c\u6cdb\u5316\u5462\uff1f\u7b54\u6848\u5c31\u662f\u90a3\u4e9b\u6b65\u957f\u8f83\u9ad8\u7684\u7279\u5f81\u3002\u56e0\u6b64\uff0c\u8fd9\u5c31\u628a\u7814\u7a76\u7126\u70b9\u5f15\u5411\u4e86\u5b66\u4e60\u7b97\u6cd5\u672c\u8eab\uff0c\u4e5f\u5c31\u662f\u7528\u4e8e\u5b66\u4e60\u6240\u6709\u6743\u91cd\u5bf9\u5e94\u6b65\u957f\u7684\u5143\u5b66\u4e60\u7b97\u6cd5\u3002<\/p>\n<p>\u5728\u6240\u6709\u7b97\u6cd5\u4e2d\uff0c\u6211\u6700\u7231\u7684\u7b97\u6cd5\u5f53\u7136\u662f\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\uff08Temporal Difference Learning\uff09\uff0c\u7b2c\u4e8c\u7231\u7684\u7b97\u6cd5\u662f\u4e13\u95e8\u7528\u6765\u8bbe\u7f6e\u6b65\u957f\u7684\uff0c\u5b83\u53eb Incremental Delta-Bar-Delta\uff0c\u7b80\u79f0 IDBD\u3002\u8fd9\u662f\u4e2a1992\u5e74\u7684\u8001\u7b97\u6cd5\u4e86\uff0c\u5b83\u7684\u7528\u9014\u975e\u5e38\u660e\u786e\uff0c\u5c31\u662f\u7528\u6765\u8bbe\u7f6e\u6b65\u957f\uff0c\u8fdb\u800c\u51b3\u5b9a\u6cdb\u5316\u662f\u5982\u4f55\u53d1\u751f\u7684\u3002\u518d\u6765\u804a\u804a RL \u4e2d\u7684\u504f\u7f6e\uff08bias\uff09\uff0c\u8fd9\u7bc7\u8bba\u6587\u63d0\u51fa\u4e86\u7c7b\u4f3c\u901a\u8fc7\u68af\u5ea6\u4e0b\u964d\uff08gradient descent\uff09\u5b66\u4e60\u504f\u7f6e\u7684\u601d\u8def\u3002\u4e5f\u5c31\u662f\u8bf4\uff0c\u6211\u4eec\u4e0d\u518d\u7531\u4eba\u5de5\u7f16\u5199\u7a0b\u5e8f\u6765\u8bbe\u5b9a\u504f\u7f6e\uff0c\u800c\u662f\u8ba9\u7cfb\u7edf\u81ea\u4e3b\u5b66\u4e60\u5982\u4f55\u6cdb\u5316\uff0c\u8fd9\u672c\u8d28\u4e0a\u5c31\u662f\u5728\u5b66\u4e60\u504f\u7f6e\u3002IDBD \u7b97\u6cd5\u867d\u7136\u662f\u4e2a\u8001\u7b97\u6cd5\uff0c\u4f46\u5b83\u5df2\u7ecf\u8fc7\u4e00\u4e9b\u6539\u8fdb\uff0c\u6211\u4eec\u76ee\u524d\u4e5f\u5728\u5bf9\u5b83\u8fdb\u884c\u66f4\u65b0\uff0c\u8fd8\u5728\u5c1d\u8bd5\u5c06\u5176\u6269\u5c55\u5230\u6df1\u5ea6\u5b66\u4e60\u9886\u57df\uff0c\u53ea\u662f\u8fd9\u9879\u5de5\u4f5c\u4ecd\u5728\u8fdb\u884c\u4e2d\uff0c\u5b83\u6709\u5f88\u5927\u7684\u6f5c\u529b\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u4e4b\u524d\u6211\u4eec\u804a\u5230\u8fc7\u5b66\u4e60\u6548\u7387\uff0c\u4f60\u4e5f\u8bf4\u8fc7\u6548\u7387\u5f53\u7136\u662f\u8d8a\u9ad8\u8d8a\u597d\uff0c\u800c\u4e14\u8fd9\u662f\u673a\u5668\u5b66\u4e60\u9886\u57df\u7684\u666e\u904d\u95ee\u9898\uff0c\u6240\u4ee5\u68af\u5ea6\u4e0b\u964d\u5728\u8fc7\u53bb\u786e\u5b9e\u63a8\u52a8\u4e86\u4eba\u5de5\u667a\u80fd\u7684\u53d1\u5c55\u3002\u4f46\u53e6\u4e00\u65b9\u9762\uff0c\u5b83\u6216\u8bb8\u5e76\u4e0d\u662f\u5b66\u4e60\u795e\u7ecf\u7f51\u7edc\u51fd\u6570\u6620\u5c04\u7684\u9ad8\u6548\u65b9\u6cd5\u3002\u4f60\u8ba4\u4e3a\u8fd9\u662f\u6211\u4eec\u9700\u8981\u7a81\u7834\u7684\u74f6\u9888\u5417\uff1f\u8fd8\u662f\u8bf4\uff0c\u4f60\u89c9\u5f97\u68af\u5ea6\u4e0b\u964d\u4ecd\u662f\u672a\u6765\u7684\u53d1\u5c55\u65b9\u5411\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u5982\u679c\u6211\u4eec\u8bd5\u7740\u63d0\u70bc\u667a\u80fd\u4e0e\u5b66\u4e60\u7684\u51e0\u6761\u57fa\u672c\u539f\u5219\uff0c<strong>\u68af\u5ea6\u4e0b\u964d\u4e00\u5b9a\u662f\u5176\u4e2d\u4e4b\u4e00<\/strong>\uff0c\u5b83\u7684\u5a01\u529b\u5f3a\u5927\u4e14\u81f3\u5173\u91cd\u8981\u3002\u95ee\u9898\u5728\u4e8e\uff0c\u5f53\u6211\u4eec\u719f\u7ec3\u638c\u63e1\u68af\u5ea6\u4e0b\u964d\u540e\uff0c\u5c31\u5f00\u59cb\u89c9\u5f97\u5b83\u65e0\u6bd4\u51fa\u8272\uff0c\u751a\u81f3\u53ef\u80fd\u8ba4\u4e3a\u4e00\u5207\u5b66\u4e60\u95ee\u9898\u90fd\u80fd\u7528\u68af\u5ea6\u4e0b\u964d\u89e3\u51b3\u3002\u800c\u8fd9\u6070\u6070\u662f\u9519\u8bef\u6240\u5728\uff0c\u6211\u4eec\u9677\u5165\u4e86\u53ea\u7528\u68af\u5ea6\u4e0b\u964d\uff0c\u800c\u6392\u9664\u6240\u6709\u5176\u4ed6\u53ef\u80fd\u6027\u7684\u5c40\u9650\u4e2d\u3002<\/p>\n<p>\u6240\u4ee5\u5982\u679c\u8981\u5217\u4e3e\u667a\u80fd\u7684\u57fa\u672c\u539f\u5219\uff0c\u6211\u8ba4\u4e3a<strong>\u68af\u5ea6\u4e0b\u964d\u3001\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u3001\u641c\u7d22\u90fd\u662f\u5176\u4e2d\u4e4b\u4e00\uff0c\u6b64\u5916\u8fd8\u6709\u5176\u4ed6\u539f\u5219\uff0c\u4f46\u4e0d\u4f1a\u592a\u591a\u4e86<\/strong>\u3002\u800c\u641c\u7d22\u5c24\u5176\u91cd\u8981\uff0c\u5176\u5b9e\u641c\u7d22\u7684\u5f71\u5b50\u5df2\u7ecf\u5b58\u5728\u4e8e\u53cd\u5411\u4f20\u64ad\u4e2d\u4e86\uff0c\u8bad\u7ec3\u5f00\u59cb\u65f6\uff0c\u6211\u4eec\u4f1a\u7ed9\u6a21\u578b\u8bbe\u7f6e\u968f\u673a\u7684\u5c0f\u6743\u91cd\uff0c\u4f46\u4e5f\u4ec5\u4ec5\u662f\u5728\u5f00\u59cb\u65f6\u8fd9\u6837\u505a\u3002\u4ece\u90a3\u4e4b\u540e\uff0c\u6a21\u578b\u5c31\u53ea\u4f1a\u5b8c\u5168\u786e\u5b9a\u6027\u5730\u6cbf\u7740\u68af\u5ea6\u65b9\u5411\u66f4\u65b0\u3002<\/p>\n<p>\u5728\u8fc7\u53bb\u51e0\u5e74\u91cc\uff0c\u6211\u4eec\u5728\u6301\u7eed\u5b66\u4e60\u9886\u57df\u53d6\u5f97\u4e86\u4e00\u70b9\u7a81\u7834\uff0c\u6211\u4eec\u5bf9\u53cd\u5411\u4f20\u64ad\u8fdb\u884c\u4e86\u6539\u8fdb\uff0c\u63d0\u51fa\u4e86\u6301\u7eed\u53cd\u5411\u4f20\u64ad\u3002\u8fd9\u79cd\u65b9\u6cd5\u4e0b\uff0c\u6a21\u578b\u4e0d\u518d\u662f\u4e00\u76f4\u53ea\u6cbf\u7740\u68af\u5ea6\u66f4\u65b0\uff0c\u4e5f\u4e0d\u518d\u53ea\u5728\u521d\u59cb\u9636\u6bb5\u8bbe\u7f6e\u968f\u673a\u6743\u91cd\uff0c\u800c\u662f\u5728\u6574\u4e2a\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u6301\u7eed\u91cd\u65b0\u6ce8\u5165\u968f\u673a\u6743\u91cd\u3002\u4e5f\u5c31\u662f\u8bf4\uff0c\u968f\u673a\u5316\u4e0d\u518d\u53ea\u53d1\u751f\u5728\u5f00\u5934\uff0c\u800c\u662f\u8d2f\u7a7f\u59cb\u7ec8\uff0c\u8fd9\u6837\u5c31\u662f\u5b8c\u5168\u8fde\u7eed\u7684\u3002\u968f\u673a\u6743\u91cd\u6b63\u662f\u641c\u7d22\u7684\u6838\u5fc3\u8981\u7d20\uff0c\u5b83\u80fd\u5c06\u6a21\u578b\u63a8\u5411\u968f\u673a\u7684\u65b9\u5411\uff0c\u4ece\u800c\u63a2\u7d22\u66f4\u591a\u53ef\u80fd\u6027\u3002\u56e0\u6b64\uff0c\u6211\u4eec\u7684\u68af\u5ea6\u4e0b\u964d\u7b97\u6cd5\u4e2d\u9700\u8981\u878d\u5165\u8fd9\u79cd\u641c\u7d22\u6210\u5206\uff0c\u628a\u641c\u7d22\u548c\u68af\u5ea6\u4e0b\u964d\u7ed3\u5408\u8d77\u6765\uff0c\u5176\u5a01\u529b\u4f1a\u8fdc\u5927\u4e8e\u68af\u5ea6\u4e0b\u964d\u5355\u72ec\u4f7f\u7528\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u4f60\u4e4b\u524d\u63d0\u5230\u4e86\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\uff0c\u4f60\u662f\u5982\u4f55\u53d1\u660e\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u7684\uff1f\u8fd9\u79cd\u51b3\u7b56\u2014\u884c\u52a8\u7684\u903b\u8f91\u4e0e\u6574\u4e2a\u63a7\u5236\u9886\u57df\u606f\u606f\u76f8\u5173\uff0c\u4ed6\u4eec\u4e00\u76f4\u5728\u7814\u7a76\u8d1d\u5c14\u66fc\u65b9\u7a0b\u4f18\u5316\u7b49\u95ee\u9898\uff0c\u800c RL \u6b63\u662f\u4ece\u8fd9\u4e9b\u57fa\u7840\u4e0a\u53d1\u5c55\u800c\u6765\u7684\uff0c\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u662f\u5176\u4e2d\u7684\u5173\u952e\u3002\u8fd8\u6709\u6211\u4e4b\u524d\u63d0\u5230\u8fc7\uff0c\u4f60\u7684\u4e00\u7bc7\u65e9\u671f\u8bba\u6587\u603b\u662f\u88ab\u6211\u7528\u4f5c\u6848\u4f8b\uff0c\u8fd9\u7bc7\u8bba\u6587\u4ece\u5fc3\u7406\u5b66\u4e2d\u7684\u7ecf\u5178\u6761\u4ef6\u53cd\u5c04\u3001\u795e\u7ecf\u79d1\u5b66\u7b49\u9886\u57df\u5207\u5165\uff0c\u7136\u540e\u5c06\u601d\u8def\u5ef6\u4f38\u5230\u4e86\u8ba1\u7b97\u673a\u79d1\u5b66\u9886\u57df\uff0c\u8fd9\u7bc7\u8bba\u6587\u975e\u5e38\u51fa\u8272\uff0c\u80fd\u548c\u6211\u4eec\u591a\u8bb2\u8bb2\u8fd9\u7bc7\u8bba\u6587\u4ee5\u53ca\u80cc\u540e\u7684\u6545\u4e8b\u5417\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u7684\u786e\u6e90\u4e8e\u7ecf\u5178\u6761\u4ef6\u53cd\u5c04\uff0c\u5b83\u8131\u80ce\u4e8e\u5fc3\u7406\u5b66\uff0c\u4ee5\u53ca\u4e0a\u4e16\u7eaa\u65e9\u671f\u6240\u6709\u5173\u4e8e\u52a8\u7269\u5b66\u4e60\u7684\u5148\u9a71\u6027\u5b9e\u9a8c\u3002\u7ecf\u5178\u6761\u4ef6\u53cd\u5c04\u662f\u5df4\u752b\u6d1b\u592b\u5728 19 \u4e16\u7eaa 90 \u5e74\u4ee3\u5de6\u53f3\u63d0\u51fa\u7684\u6982\u5ff5\u3002\u6211\u672c\u79d1\u65f6\u5b66\u7684\u5c31\u662f\u5fc3\u7406\u5b66\uff0c\u4f46\u6211\u8fd8\u662f\u4e00\u76f4\u90fd\u5728\u7814\u7a76\u4eba\u5de5\u667a\u80fd\uff0c\u53ea\u662f\u5728 70 \u5e74\u4ee3\uff0c\u672c\u79d1\u9636\u6bb5\u8fd8\u6ca1\u6cd5\u4e3b\u4fee\u8ba1\u7b97\u673a\u79d1\u5b66\uff0c\u5fc5\u987b\u9009\u5176\u4ed6\u4e13\u4e1a\uff0c\u6211\u5f88\u4e50\u610f\u4e3b\u4fee\u5fc3\u7406\u5b66\uff0c\u56e0\u4e3a\u5728\u5b66\u4e60\u9886\u57df\uff0c\u5c24\u5176\u662f\u5728\u63a2\u7d22\u5b66\u4e60\u7684\u57fa\u672c\u539f\u7406\u65b9\u9762\uff0c\u5fc3\u7406\u5b66\u5bb6\u7684\u601d\u8003\u662f\u6700\u6df1\u5165\u7684\uff0c\u4ed6\u4eec\u4e3a\u6b64\u4ed8\u51fa\u4e86\u5de8\u5927\u52aa\u529b\uff0c\u63d0\u51fa\u4e86\u8bb8\u591a\u91cd\u8981\u7684\u89c2\u70b9\uff0c\u5176\u4e2d\u6709\u4e00\u4e9b\u89c2\u70b9\u867d\u7136\u6ca1\u6709\u76f4\u63a5\u50ac\u751f\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\uff0c\u4f46\u5df2\u7ecf\u975e\u5e38\u63a5\u8fd1\u4e86\u3002<strong>\u5982\u679c\u4f60\u4ed4\u7ec6\u68b3\u7406\u6240\u6709\u7684\u52a8\u7269\u5b9e\u9a8c\uff0c\u8bd5\u7740\u4ece\u4e2d\u63d0\u70bc\u89c4\u5f8b\uff0c\u5c31\u5fc5\u7136\u4f1a\u8d70\u5411\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\uff0c\u5b83\u5c31\u662f\u8fd9\u6837\u8bde\u751f\u7684\u3002<\/strong><\/p>\n<p>\u76f4\u5230\u540e\u6765\u6211\u4eec\u624d\u610f\u8bc6\u5230\uff0c\u4ece\u52a8\u6001\u89c4\u5212\u548c\u63a7\u5236\u7406\u8bba\u7684\u89d2\u5ea6\u6765\u770b\uff0c\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u540c\u6837\u8bf4\u5f97\u901a\uff0c\u4f46\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u6700\u521d\u662f\u6e90\u4e8e\u7ecf\u9a8c\u7684\u4ea7\u7269\u3002\u53ef\u4ee5\u8fd9\u6837\u7406\u89e3\u4e8c\u8005\u7684\u5173\u7cfb\uff1a<strong>\u52a8\u6001\u89c4\u5212\u7684\u524d\u63d0\u662f\uff0c\u4f60\u5df2\u7ecf\u5b8c\u5168\u4e86\u89e3\u6574\u4e2a\u4e16\u754c\uff0c\u4f60\u5c31\u80fd\u8ba1\u7b97\u51fa\u6700\u4f18\u89e3\uff1b\u800c\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u5219\u4e0d\u540c\uff0c\u5373\u4fbf\u6211\u4e0d\u77e5\u9053\u4e16\u754c\u662f\u5982\u4f55\u8fd0\u4f5c\u7684\uff0c\u6211\u4e5f\u80fd\u901a\u8fc7\u7ecf\u9a8c\u800c\u975e\u77e5\u8bc6\u8fbe\u6210\u4e0e\u52a8\u6001\u89c4\u5212\u76f8\u540c\u7684\u76ee\u6807\u3002<\/strong><\/p>\n<p>\u8fd9\u4e00\u70b9\u592a\u7f8e\u5999\u4e86\uff0c\u56e0\u4e3a\u5b83\u610f\u5473\u7740\u6240\u6709\u7814\u7a76\u667a\u80fd\u4e0e\u5b66\u4e60\u7684\u5b66\u79d1\u5b9e\u73b0\u4e86\u7edf\u4e00\uff0c\u65f6\u5e8f\u5dee\u5206\u5b66\u4e60\u7684\u5f71\u54cd\u529b\u975e\u5e38\u6df1\u8fdc\u3002\u56de\u5230\u795e\u7ecf\u79d1\u5b66\u9886\u57df\uff0c\u5b83\u4e3a\u7814\u7a76\u8005\u4eec\u89c2\u5bdf\u5230\u7684\u8bb8\u591a\u52a8\u7269\u5b66\u4e60\u65b0\u7ed3\u679c\u3001\u65b0\u6570\u636e\u63d0\u4f9b\u4e86\u5408\u7406\u89e3\u91ca\uff0c\u5982\u4eca\u5df2\u6210\u4e3a\u5927\u8111\u5956\u52b1\u7cfb\u7edf\u7814\u7a76\u4e2d\u5360\u4e3b\u5bfc\u5730\u4f4d\u7684\u6807\u51c6\u6a21\u578b\u4e0e\u7406\u8bba\u3002\u6211\u4eec\u80fd\u770b\u5230\uff0c\u4eba\u5de5\u667a\u80fd\u3001\u63a7\u5236\u7406\u8bba\u3001\u795e\u7ecf\u79d1\u5b66\u8fd9\u4e9b\u539f\u672c\u622a\u7136\u4e0d\u540c\u7684\u5b66\u79d1\uff0c\u6b63\u6210\u529f\u5730\u8d70\u5411\u878d\u5408\u4e0e\u4ea4\u6c47\u3002\u6211\u5bf9\u6b64\u611f\u5230\u975e\u5e38\u6b23\u6170\uff0c\u8fd9\u5c31\u50cf\u7ec8\u4e8e\u89e6\u78b0\u5230\u4e86\u667a\u80fd\u6700\u6839\u672c\u7684\u4e1c\u897f\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u6211\u6ce8\u610f\u5230\u7684\u4e00\u70b9\u533a\u522b\u4e5f\u5c31\u662f\u65e0\u6a21\u578b\uff08model-free\uff09\u548c\u57fa\u4e8e\u6a21\u578b\uff08model-based\uff09\u7684\u6982\u5ff5\uff0c\u4f60\u80fd\u518d\u8be6\u7ec6\u89e3\u91ca\u4e00\u4e0b\u8fd9\u4e2a\u5417\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u9996\u5148\uff0c\u6211\u4eec\u5f97\u660e\u786e\u4e00\u70b9\uff0c\u6211\u4eec\u4e24\u8005\u90fd\u9700\u8981\u517c\u987e\u3002<strong>\u6211\u4eec\u65e2\u5e0c\u671b\u76f4\u63a5\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\uff0c\u4e5f\u5e0c\u671b\u95f4\u63a5\u4ece\u7ecf\u9a8c\u4e2d\u5b66\u4e60\uff0c\u6211\u4eec\u5e0c\u671b\u81ea\u5df1\u7684\u7ecf\u9a8c\u80fd\u5f62\u6210\u4e00\u4e2a\u4e16\u754c\u6a21\u578b\uff0c\u7136\u540e\u5229\u7528\u8fd9\u4e2a\u6a21\u578b\u6765\u89c4\u5212\u884c\u4e3a\u4e0a\u7684\u6539\u8fdb\u3002<\/strong>\u8fd9\u79cd\u89c4\u5212\u5c31\u7c7b\u4f3c\u4e8e\u63a8\u7406\uff0c\u6211\u4eec\u4ec5\u51ed\u5bf9\u4e16\u754c\u7684\u8ba4\u77e5\u5c31\u80fd\u60f3\u6e05\u695a\u8be5\u505a\u4ec0\u4e48\u3002\u8fd9\u4e24\u8005\u6211\u4eec\u90fd\u9700\u8981\uff0c\u800c\u5b9e\u73b0\u4e8c\u8005\u5728\u67d0\u79cd\u610f\u4e49\u4e0a\u7684\u7edf\u4e00\uff0c\u6b63\u662f\u6211\u4eec\u8ffd\u6c42\u7684\u201cThe Holy Grail\u201d\uff08\u5723\u676f\uff09\uff0c\u5c31\u662f\u80fd\u591f\u540c\u65f6\u8fdb\u884c\u89c4\u5212\u548c\u5b66\u4e60\uff0c\u5e76\u5728\u67d0\u79cd\u610f\u4e49\u4e0a\u5c06\u5b83\u4eec\u7edf\u4e00\u8d77\u6765\u3002<\/p>\n<p>\u8fd9\u79cd\u7edf\u4e00\u7684\u5f00\u7aef\u662f\u201cdyna system\u201d\uff08\u52a8\u529b\u7cfb\u7edf\uff09\uff0c\u5728\u8fd9\u4e2a\u7cfb\u7edf\u4e2d\uff0c\u89c4\u5212\u7684\u8fc7\u7a0b\u672c\u8eab\u5c31\u5e26\u6709\u5b66\u4e60\u7684\u5c5e\u6027\uff0c\u4e0d\u8fc7\u8fd9\u79cd\u5b66\u4e60\u662f\u901a\u8fc7\u60f3\u8c61\u81ea\u5df1\u53ef\u80fd\u91c7\u53d6\u7684\u884c\u52a8\u6765\u5b8c\u6210\u7684\uff0c\u8fd9\u4e00\u70b9\u5728\u5fc3\u7406\u5b66\u6587\u732e\u4e2d\u4e5f\u6709\u5f88\u597d\u7684\u4f53\u73b0\uff0cRL \u7684\u76ee\u6807\u6b63\u662f\u6210\u4e3a\u4e00\u79cd\u6db5\u76d6\u6240\u6709\u667a\u80fd\u5f62\u5f0f\u7684\u5b8f\u5927\u7406\u8bba\u3002<\/p>\n<p><strong>\u6c6a\u519b<\/strong>\uff1a\u60a8\u6709\u4ec0\u4e48\u60f3\u5bf9\u4e2d\u56fd\u7684 RL \u793e\u533a\u8bf4\u7684\u5417\uff1f\u60a8\u8ba4\u4e3a\u8fd9\u4e0e\u60a8\u6b63\u5728\u5f00\u5c55\u7684 Openmind \u4f7f\u547d\u6709\u4ec0\u4e48\u8054\u7cfb\uff1f<\/p>\n<p><strong>Richard Sutton<\/strong>\uff1a\u6211\u60f3\u8865\u5145\u4e00\u70b9\u6211\u4eec\u5c1a\u672a\u8c08\u53ca\u7684\u6838\u5fc3\u5185\u5bb9\uff0c\u5f97\u76ca\u4e8e\u6469\u5c14\u5b9a\u5f8b\u4ee5\u53ca\u8ba1\u7b97\u80fd\u529b\u7684\u65e5\u76ca\u666e\u53ca\uff0c\u5f53\u4e0b\u6b63\u662f\u5b87\u5b99\u4e2d\u4e00\u4e2a\u6781\u5176\u7279\u6b8a\u7684\u65f6\u4ee3\uff0c\u5728\u672a\u6765\u7684\u4e00\u4e8c\u5341\u5e74\u91cc\uff0c\u6211\u4eec\u5c06\u5f97\u4ee5\u5f04\u660e\u767d\u601d\u7ef4\u662f\u5982\u4f55\u8fd0\u4f5c\u7684\u3002\u80fd\u751f\u6d3b\u5728\u8fd9\u6837\u4e00\u4e2a\u65f6\u4ee3\uff0c\u5b9e\u5728\u592a\u4ee4\u4eba\u60ca\u53f9\u4e86\u3002\u6211\u66fe\u5c06\u5bf9\u667a\u80fd\u7684\u7406\u89e3\u6bd4\u4f5c\u5730\u7403\u4e0a\u751f\u547d\u7684\u8d77\u6e90\uff0c\u6211\u8ba4\u4e3a\u8fd9\u662f\u4e00\u4ef6\u5177\u6709\u540c\u7b49\u91cd\u5927\u610f\u4e49\u7684\u4e8b\u4ef6\u3002\u6211\u4eec\u5c06\u80fd\u591f\u8bbe\u8ba1\u51fa\u4e0e\u4eba\u7c7b\u81ea\u8eab\u4e00\u6837\u5f3a\u5927\u3001\u751a\u81f3\u53ef\u80fd\u66f4\u5f3a\u5927\u7684 agent\uff0c\u8fd9\u771f\u7684\u662f\u4e00\u4ef6\u65e0\u6bd4\u6fc0\u52a8\u4eba\u5fc3\u7684\u5927\u4e8b\u3002\u5982\u679c\u8bf4\u6211\u6700\u60f3\u8868\u8fbe\u4ec0\u4e48\uff0c\u90a3\u5c31\u662f\u6211\u5bf9\u6b64\u62b1\u6709\u6781\u5927\u7684\u4e50\u89c2\uff0c\u6211\u8ba4\u4e3a\u4e00\u5207\u90fd\u4f1a\u53d8\u5f97\u975e\u5e38\u597d\uff0c\u8fd9\u4e0d\u4ec5\u5c06\u5b9e\u73b0\u5386\u53f2\u4e0a\u77e5\u8bc6\u5206\u5b50\u4e43\u81f3\u666e\u901a\u4eba\u4e00\u76f4\u4ee5\u6765\u8ffd\u5bfb\u7684\u4f1f\u5927\u63a2\u7d22\u76ee\u6807\uff0c\u8fd8\u5c06\u5bf9\u4e16\u754c\u7684\u7ecf\u6d4e\u53d1\u5c55\u4ea7\u751f\u5de8\u5927\u7684\u79ef\u6781\u5f71\u54cd\u3002<\/p>\n<p>\u672a\u6765\u5e76\u975e\u5168\u90fd\u662f\u4e00\u5e06\u98ce\u987a\u7684\uff0c\u80af\u5b9a\u4f1a\u9762\u4e34\u8bf8\u591a\u6311\u6218\uff0c\u4f46\u8fd9\u5e76\u975e\u56e0\u4e3a\u4eba\u5de5\u667a\u80fd\u4f1a\u5e26\u6765\u4ec0\u4e48\u7cdf\u7cd5\u7684\u7ed3\u679c\uff0c\u603b\u7684\u6765\u8bf4\uff0c\u5b83\u662f\u4e00\u4ef6\u6781\u5177\u79ef\u6781\u610f\u4e49\u7684\u4e8b\u3002\u5b83\u5c06\u4e3a\u6211\u4eec\u5e26\u6765\u66f4\u6df1\u5165\u7684\u79d1\u5b66\u7814\u7a76\u3001\u66f4\u900f\u5f7b\u7684\u81ea\u6211\u8ba4\u77e5\uff0c\u751a\u81f3\u6709\u53ef\u80fd\u5e2e\u52a9\u6211\u4eec\u66f4\u597d\u5730\u548c\u5e73\u5171\u5904\uff0c\u800c\u8fd9\u6b63\u662f Openmind \u4f7f\u547d\u7684\u4e00\u90e8\u5206\u3002<\/p>\n<p>\u6bcf\u4e2a\u4eba\u90fd\u80fd\u53c2\u4e0e\u5230\u8fd9\u4efd\u7406\u89e3\u4e2d\u6765\uff0c\u6211\u4eec\u4f1a\u5c55\u5f00\u5408\u4f5c\uff0c\u5171\u540c\u53bb\u63a2\u7d22\u4eba\u5de5\u667a\u80fd\u7684\u5965\u79d8\uff0c\u5171\u540c\u6536\u83b7\u5b83\u5e26\u6765\u7684\u6210\u679c\uff0c\u5171\u540c\u52a0\u6df1\u5bf9\u81ea\u6211\u7684\u8ba4\u77e5\uff0c\u800c\u4e14\u8981\u660e\u767d\uff0c\u5e76\u975e\u53ea\u6709\u673a\u5668\u4f1a\u53d8\u5f97\u667a\u80fd\uff0c\u6211\u4eec\u4eba\u7c7b\u81ea\u8eab\u4e5f\u4f1a\u901a\u8fc7\u589e\u5f3a\u624b\u6bb5\u53d8\u5f97\u66f4\u806a\u660e\u3002\u5176\u5b9e\u6211\u4eec\u5df2\u7ecf\u5728\u88ab\u589e\u5f3a\u4e86\uff0c\u7535\u8bdd\u5728\u589e\u5f3a\u6211\u4eec\u7684\u6c9f\u901a\uff0c\u773c\u955c\u5728\u589e\u5f3a\u6211\u4eec\u7684\u89c6\u529b\uff0c\u4e00\u652f\u94c5\u7b14\u3001\u4e00\u95e8\u80fd\u8ba9\u5f7c\u6b64\u4ea4\u6d41\u7684\u8bed\u8a00\uff0c\u8fd9\u4e9b\u90fd\u662f\u5bf9\u6211\u4eec\u80fd\u529b\u7684\u589e\u5f3a\uff0c\u4eba\u5de5\u667a\u80fd\u5b9e\u9645\u4e0a\u5c31\u662f\u8fd9\u4e00\u589e\u5f3a\u8fc7\u7a0b\u7684\u5ef6\u7eed\uff0c\u6240\u4ee5\u4eba\u5de5\u667a\u80fd\u7684\u53d1\u5c55\u524d\u666f\u4f1a\u975e\u5e38\u597d\u3002\u6211\u4eec\u5e94\u5f53\u4e3a\u4ece\u4e8b\u8fd9\u9879\u5de5\u4f5c\u800c\u611f\u5230\u7531\u8877\u7684\u559c\u60a6\uff0c\u5e76\u4e3a\u4e86\u5168\u4eba\u7c7b\u7684\u798f\u7949\u53bb\u63a8\u8fdb\u5b83\u7684\u53d1\u5c55\u3002<\/p>\n<p>\uff08\u96f7\u5cf0\u7f51<span style=\"color:#FD5D3C\">(\u516c\u4f17\u53f7\uff1a\u96f7\u5cf0\u7f51)<\/span>\uff09<\/p>\n<p>\u96f7\u5cf0\u7f51\u539f\u521b\u6587\u7ae0\uff0c\u672a\u7ecf\u6388\u6743\u7981\u6b62\u8f6c\u8f7d\u3002\u8be6\u60c5\u89c1<a href=\"https:\/\/r.xiumi.us\/board\/v5\/3qhbI\/392977584\" rel=\"nofollow\" target=\"_blank\">\u8f6c\u8f7d\u987b\u77e5<\/a>\u3002<\/p>\n<p>\n                        <a href=\"https:\/\/vip.yanxishe.com\/?from=leiphonearticle\" target=\"_blank\"><br \/>\n                            <a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-ef1b38dfb738ad25d47c69ef4a8cea34.jpeg\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-ef1b38dfb738ad25d47c69ef4a8cea34.jpeg\"><\/a><br \/>\n                        <\/a>\n                    <\/p>\n<p>                                            \u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:<a href=\"https:\/\/www.leiphone.com\/category\/ai\/2roeBEayl5wf0mne.html\" target=\"_blank\">\u6c6a\u519b\u5bf9\u8bdd Rich Sutton\uff1a\u5927\u6a21\u578b\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u6211\u4eec\u5bf9\u667a\u80fd\u7406\u89e3\u7684\u6ce8\u610f\u529b<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u5728\u521a\u521a\u843d\u5e55\u7684 RL China 2025 [&hellip;]<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[27],"tags":[70,68,55,65],"class_list":["post-40480","post","type-post","status-publish","format-standard","hentry","category-news","tag-agent","tag-68","tag-55","tag-65"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v26.4 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u6c6a\u519b\u5bf9\u8bdd Rich Sutton\uff1a\u5927\u6a21\u578b\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u6211\u4eec\u5bf9\u667a\u80fd\u7406\u89e3\u7684\u6ce8\u610f\u529b - \u4e00\u8d77AI\u6280\u672f<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/17aitech.com\/?p=40480\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/17aitech.com\/?p=40480\",\"url\":\"https:\/\/17aitech.com\/?p=40480\",\"name\":\"\u6c6a\u519b\u5bf9\u8bdd Rich Sutton\uff1a\u5927\u6a21\u578b\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u6211\u4eec\u5bf9\u667a\u80fd\u7406\u89e3\u7684\u6ce8\u610f\u529b - \u4e00\u8d77AI\u6280\u672f\",\"isPartOf\":{\"@id\":\"https:\/\/17aitech.com\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/17aitech.com\/?p=40480#primaryimage\"},\"image\":{\"@id\":\"https:\/\/17aitech.com\/?p=40480#primaryimage\"},\"thumbnailUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-6d838055d4078cf6c79e19d40ae1c90d.png\",\"datePublished\":\"2025-09-28T14:05:16+00:00\",\"author\":{\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d\"},\"breadcrumb\":{\"@id\":\"https:\/\/17aitech.com\/?p=40480#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/17aitech.com\/?p=40480\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/?p=40480#primaryimage\",\"url\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-6d838055d4078cf6c79e19d40ae1c90d.png\",\"contentUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-6d838055d4078cf6c79e19d40ae1c90d.png\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/17aitech.com\/?p=40480#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/17aitech.com\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u6c6a\u519b\u5bf9\u8bdd Rich Sutton\uff1a\u5927\u6a21\u578b\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u6211\u4eec\u5bf9\u667a\u80fd\u7406\u89e3\u7684\u6ce8\u610f\u529b\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/17aitech.com\/#website\",\"url\":\"https:\/\/17aitech.com\/\",\"name\":\"\u4e00\u8d77AI\u6280\u672f\",\"description\":\"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca\",\"alternateName\":\"\u4e00\u8d77AI\u6280\u672f\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/17aitech.com\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d\",\"name\":\"AI\u5c0f\u52a9\u624b\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/image\/\",\"url\":\"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png\",\"contentUrl\":\"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png\",\"caption\":\"AI\u5c0f\u52a9\u624b\"},\"description\":\"\u8fd9\u4e2a\u4eba\u5f88\u61d2\uff0c\u4ec0\u4e48\u90fd\u6ca1\u6709\u7559\u4e0b\uff5e\",\"url\":\"https:\/\/17aitech.com\/?page_id=33738&user=3\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u6c6a\u519b\u5bf9\u8bdd Rich Sutton\uff1a\u5927\u6a21\u578b\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u6211\u4eec\u5bf9\u667a\u80fd\u7406\u89e3\u7684\u6ce8\u610f\u529b - \u4e00\u8d77AI\u6280\u672f","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/17aitech.com\/?p=40480","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/17aitech.com\/?p=40480","url":"https:\/\/17aitech.com\/?p=40480","name":"\u6c6a\u519b\u5bf9\u8bdd Rich Sutton\uff1a\u5927\u6a21\u578b\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u6211\u4eec\u5bf9\u667a\u80fd\u7406\u89e3\u7684\u6ce8\u610f\u529b - \u4e00\u8d77AI\u6280\u672f","isPartOf":{"@id":"https:\/\/17aitech.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/17aitech.com\/?p=40480#primaryimage"},"image":{"@id":"https:\/\/17aitech.com\/?p=40480#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-6d838055d4078cf6c79e19d40ae1c90d.png","datePublished":"2025-09-28T14:05:16+00:00","author":{"@id":"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d"},"breadcrumb":{"@id":"https:\/\/17aitech.com\/?p=40480#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/17aitech.com\/?p=40480"]}]},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/?p=40480#primaryimage","url":"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-6d838055d4078cf6c79e19d40ae1c90d.png","contentUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/09\/frc-6d838055d4078cf6c79e19d40ae1c90d.png"},{"@type":"BreadcrumbList","@id":"https:\/\/17aitech.com\/?p=40480#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/17aitech.com\/"},{"@type":"ListItem","position":2,"name":"\u6c6a\u519b\u5bf9\u8bdd Rich Sutton\uff1a\u5927\u6a21\u578b\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u5206\u6563\u4e86\u6211\u4eec\u5bf9\u667a\u80fd\u7406\u89e3\u7684\u6ce8\u610f\u529b"}]},{"@type":"WebSite","@id":"https:\/\/17aitech.com\/#website","url":"https:\/\/17aitech.com\/","name":"\u4e00\u8d77AI\u6280\u672f","description":"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca","alternateName":"\u4e00\u8d77AI\u6280\u672f","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/17aitech.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d","name":"AI\u5c0f\u52a9\u624b","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/#\/schema\/person\/image\/","url":"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png","contentUrl":"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png","caption":"AI\u5c0f\u52a9\u624b"},"description":"\u8fd9\u4e2a\u4eba\u5f88\u61d2\uff0c\u4ec0\u4e48\u90fd\u6ca1\u6709\u7559\u4e0b\uff5e","url":"https:\/\/17aitech.com\/?page_id=33738&user=3"}]}},"_links":{"self":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/40480","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=40480"}],"version-history":[{"count":0,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/40480\/revisions"}],"wp:attachment":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=40480"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=40480"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=40480"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}