{"id":46779,"date":"2023-11-27T22:44:37","date_gmt":"2023-11-27T14:44:37","guid":{"rendered":"https:\/\/swarma.org\/?p=46779"},"modified":"2023-11-27T22:44:37","modified_gmt":"2023-11-27T14:44:37","slug":"%e5%89%8d%e6%b2%bf%e8%bf%9b%e5%b1%95%ef%bc%9a%e5%9c%a8%e7%bb%84%e5%90%88%e4%bc%98%e5%8c%96%e9%97%ae%e9%a2%98%e4%b8%ad%e9%87%8d%e6%96%b0%e5%ae%a1%e8%a7%86%e9%87%87%e6%a0%b7","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=46779","title":{"rendered":"\u524d\u6cbf\u8fdb\u5c55\uff1a\u5728\u7ec4\u5408\u4f18\u5316\u95ee\u9898\u4e2d\u91cd\u65b0\u5ba1\u89c6\u91c7\u6837"},"content":{"rendered":"
\n
\n
\n
\n

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

\n
\n
\n
\n
\n
\n

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

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

\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\n
\n
\u5728\u96c6\u667a\u4ff1\u4e50\u90e8\u300c\u56fe\u795e\u7ecf\u7f51\u7edc\u4e0e\u7ec4\u5408\u4f18\u5316\u8bfb\u4e66\u4f1a<\/a>\u300d\u7b2c\u4e00\u5b63\u6700\u540e\u4e00\u671f\uff0cDeepMind \u5b9e\u9a8c\u5ba4\u8d44\u6df1\u79d1\u5b66\u5bb6\u6234\u6db5\u4fca\u9996\u5148\u4ecb\u7ecd\u4e86\u4ed6\u4eec\u8fd1\u671f\u53d1\u8868\u4e8e PMLR<\/em> \u7684\u4e00\u9879\u6700\u65b0\u7814\u7a76\uff0c\u201c\u5728\u7ec4\u5408\u4f18\u5316\u95ee\u9898\u4e2d\u91cd\u65b0\u5ba1\u89c6\u62bd\u6837\u201d\uff08Revisiting Sampling for Combinatorial Optimization\uff09\u3002<\/span><\/strong><\/section>\n
<\/span><\/strong><\/section>\n<\/section>\n
\n
<\/span><\/strong><\/section>\n
<\/span><\/strong><\/section>\n<\/section>\n
\n
<\/span><\/section>\n

<\/section>\n
\u7814\u7a76\u9886\u57df\uff1a\u56fe\u795e\u7ecf\u7f51\u7edc\uff0c\u7ec4\u5408\u4f18\u5316\uff0c\u91c7\u6837\u7b97\u6cd5\uff0c\u79bb\u6563\u7a7a\u95f4<\/strong><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/section>\n<\/section>\n
\n
<\/span><\/strong><\/section>\n<\/section>\n
\n
<\/span><\/strong><\/strong><\/strong><\/span><\/strong><\/strong><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/strong><\/strong><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\n
\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\n
<\/section>\n<\/section>\n
\n
<\/span><\/strong><\/section>\n
\u6234\u6db5\u4fca<\/span> <\/span>| \u8bb2\u8005<\/span><\/strong><\/span><\/strong><\/section>\n
<\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/section>\n
\u5218\u660e\u660a<\/span> <\/span>| \u6574\u7406<\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/span><\/strong><\/section>\n
<\/span><\/strong><\/section>\n

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


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

\n
\n
\n

\u76ee\u5f55<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n

    \n
      \n
    • \n

      \u91c7\u6837\u7b97\u6cd5\u6982\u8ff0<\/p>\n<\/li>\n

    • \n

      \u9ad8\u6548\u7684\u79bb\u6563\u7a7a\u95f4\u91c7\u6837\u7b97\u6cd5<\/p>\n<\/li>\n

    • \n

      \u91c7\u6837\u7b97\u6cd5\u4e0e\u7ec4\u5408\u4f18\u5316<\/p>\n<\/li>\n

    • \n

      \u91c7\u6837\u7b97\u6cd5\u4e0eAI\u6a21\u578b<\/p>\n<\/li>\n<\/ul>\n<\/ul>\n


      <\/section>\n

      <\/section>\n

      <\/section>\n

      \n
      \n
      \n
      \n
      \n

      <\/section>\n

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

      1. \u91c7\u6837\u7b97\u6cd5\u6982\u8ff0 <\/span><\/strong><\/strong><\/p>\n<\/section>\n

      \n

      <\/section>\n

      <\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
       <\/span><\/section>\n
      \u91c7\u6837\u7b97\u6cd5<\/span><\/strong>\u5728\u5f88\u591a\u4efb\u52a1\u4e2d\u626e\u6f14\u7740\u5173\u952e\u7684\u89d2\u8272\u3002\u4f8b\u5982\uff0c\u5bf9\u4e8e\u6700\u8fd1\u5f88\u706b\u7684\u5927\u8bed\u8a00\u6a21\u578b\uff0c\u5176\u6838\u5fc3\u90e8\u5206\u5c31\u662f\u5728\u4ece\u81ea\u56de\u5f52\u6a21\u578b\u91cc\u9762\u8fdb\u884c\u91c7\u6837\uff0c\u5373\u5728\u7ed9\u5b9a\u4e0a\u4e0b\u6587\u7684\u60c5\u51b5\u4e0b\u5bf9\u4e2d\u95f4\u8fdb\u884c\u91c7\u6837\u3002\u57283D\u56fe\u50cf\u3001AI4Science\u7b49\u9886\u57df\uff0c\u5176\u80cc\u540e\u4e5f\u5927\u91cf\u4f9d\u8d56\u91c7\u6837\u7b97\u6cd5\u3002\u5b9e\u9645\u4e0a\uff0c\u5728\u81ea\u7136\u8bed\u8a00\u5904\u7406\u3001\u86cb\u767d\u8d28\u7ed3\u6784\u3001\u7a0b\u5e8f\u8bed\u8a00\u7b49\u9886\u57df\uff0c\u5f88\u591a\u60c5\u51b5\u4e0b\u6570\u636e\u7684\u8868\u8fbe\u5f62\u5f0f\u662f\u79bb\u6563\u548c\u7ed3\u6784\u5316\u7684\uff0c\u56e0\u6b64\u9700\u8981\u5bf9\u4e8e\u79bb\u6563\u7a7a\u95f4\u7684\u91c7\u6837\u7b97\u6cd5\u3002\u5373\u4f7f\u5bf9\u4e8e\u4e00\u4e9b\u53ef\u4ee5\u5728\u8fde\u7eed\u7a7a\u95f4\u8868\u5f81\u7684\u6570\u636e<\/span>\uff08\u4f8b\u5982\u56fe\u50cf\uff09<\/span>\uff0c\u4e5f\u53ef\u4ee5\u901a\u8fc7\u5c06\u5176\u6620\u5c04\u5230\u79bb\u6563\u7a7a\u95f4\u8fdb\u884c\u5904\u7406\uff0c\u4ece\u800c\u63d0\u5347\u8ba1\u7b97\u6548\u7387\u3002\u7136\u800c\uff0c\u8bbe\u8ba1\u5bf9\u79bb\u6563\u7a7a\u95f4\u7684\u91c7\u6837\u7b97\u6cd5\u5e76\u4e0d\u7b80\u5355\u3002\u96be\u70b9\u5728\u4e8e\u79bb\u6563\u6570\u636e\u662f\u4e0d\u53ef\u5fae\u7684\uff0c\u56e0\u6b64\u5728\u6761\u4ef6\u91c7\u6837\u6216\u975e\u81ea\u56de\u5f52\u6a21\u578b\u7b49\u60c5\u51b5\u4e0b\u5c06\u5f88\u96be\u8fdb\u884c\u91c7\u6837\u3002<\/span><\/section>\n

      <\/section>\n
      \u5728\u6b64\u4e4b\u524d\uff0c\u6709\u5fc5\u8981\u5148\u56de\u987e\u4e00\u4e0b\u5728\u8fde\u7eed\u7a7a\u95f4\u7684\u91c7\u6837\u7b97\u6cd5<\/strong>\u3002Langevin Monte Carlo\u662f\u4e00\u4e2a\u7ecf\u5178\u7684\u91c7\u6837\u7b97\u6cd5\u3002\u8003\u8651\u56fe1\u7684\u4e00\u4e2a\u98de\u673a\u76843D\u70b9\u4e91\u6a21\u578b\uff0c\u5176\u6982\u7387\u51fd\u6570\u4e3a<\/sub><\/span>\u3002\u5982\u679c\u5bf9\u5b83\u8fd0\u7528Langevin Monte Carlo\u7b97\u6cd5\u8fdb\u884c\u91c7\u6837\uff0c\u5c31\u53ef\u4ee5\u4ece\u90e8\u5206\u6570\u636e\u8fed\u4ee3\u5730\u751f\u6210\u4e00\u67b6\u5b8c\u6574\u7684\u98de\u673a\u3002\u5728\u91c7\u6837\u8fc7\u7a0b\u4e2d\uff0c\u5f88\u91cd\u8981\u7684\u4e00\u9879\u662f<\/sub><\/span>\uff0c\u4e5f\u53eb\u505a\u6253\u5206\u51fd\u6570\uff0c\u5728\u6269\u6563\u6a21\u578b\u7684\u53cd\u5411\u8fc7\u7a0b\u4e2d\u4e5f\u6709\u7c7b\u4f3c\u7684\u5e94\u7528\u3002\u4f46\u662f\u6211\u4eec\u77e5\u9053\uff0c\u5728\u79bb\u6563\u7a7a\u95f4\u4e2d\u5e76\u4e0d\u5b58\u5728\u8fd9\u6837\u7684\u68af\u5ea6\uff0c\u56e0\u6b64\u4e00\u4e2a\u5173\u952e\u95ee\u9898\u5c31\u662f\u89e3\u91ca<\/sub><\/span>\u5728\u79bb\u6563\u7a7a\u95f4\u4e2d\u7a76\u7adf\u5bf9\u5e94\u7740\u4ec0\u4e48\u3002<\/span><\/section>\n

      <\/span><\/section>\n
      <\/section>\n
      \u56fe1 \u57fa\u4e8eLangevin Monte Carlo\u7b97\u6cd5\u76843D\u70b9\u4e91\u91c7\u6837<\/span><\/section>\n

      <\/section>\n

      <\/section>\n

      \n
      \n
      \n
      \n
      \n

      <\/section>\n

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

      2. \u9ad8\u6548\u7684\u79bb\u6563\u7a7a\u95f4\u91c7\u6837\u7b97\u6cd5 <\/span><\/strong><\/strong><\/p>\n<\/section>\n

      \n

      <\/section>\n

      <\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
       <\/span><\/section>\n
      \u4f20\u7edf\u7684\u5bf9\u79bb\u6563\u7a7a\u95f4\u7684\u91c7\u6837\u7b97\u6cd5\u5305\u62ec\u968f\u673a\u6e38\u8d70\u3001Gibbs\u91c7\u6837\u7b49\uff0c\u5b83\u4eec\u672a\u80fd\u5229\u7528\u68af\u5ea6\u4fe1\u606f\uff0c\u56e0\u6b64\u975e\u5e38\u4f4e\u6548\u3002\u90a3\u4e48\uff0c\u5982\u4f55\u5c06Langevin Monte Carlo\u8fd9\u6837\u7684\u7b97\u6cd5\u5e94\u7528\u5230\u79bb\u6563\u7a7a\u95f4\u7684\u91c7\u6837\u4e2d\u5462<\/strong>\uff1f\u4e00\u4e2a\u6bd4\u8f83\u597d\u7684\u5207\u5165\u70b9\u662f\u8003\u8651\u68af\u5ea6\u4e0b\u964d\u7b97\u6cd5\uff0c\u5b83\u5b9e\u8d28\u4e0a\u662f\u5b9a\u4e49\u4e86\u4e00\u4e2a\u5fae\u5206\u65b9\u7a0b\u5f62\u5f0f\u7684\u68af\u5ea6\u6d41\uff0c\u800c\u4e0d\u540c\u7684\u68af\u5ea6\u4e0b\u964d\u7b97\u6cd5\u53d8\u4f53\uff0c\u4f8b\u5982\u7ecf\u5178\u68af\u5ea6\u4e0b\u964d\u3001\u4e34\u8fd1\u70b9\u7b97\u6cd5<\/span>\uff08proximal point algorithm\uff09<\/span>\u7b49\uff0c\u90fd\u5c5e\u4e8e\u5bf9\u8be5\u68af\u5ea6\u6d41\u7684\u8fd1\u4f3c\u6a21\u62df\u3002\u6839\u636e\u6587\u732e\uff0cLangevin Monte Carlo\u7b97\u6cd5\u4e5f\u53ef\u4ee5\u6309\u7167\u7c7b\u4f3c\u7684\u601d\u8def\u8fdb\u884c\u7406\u89e3\u3002\u5982\u56fe2\uff0c\u5b83\u5b9e\u8d28\u4e0a\u4e5f\u662f\u5bf9\u4e8eLangevin Dynamics\u8fd9\u4e00\u5fae\u5206\u65b9\u7a0b\u7684\u4e00\u79cd\u6a21\u62df\u3002\u56e0\u6b64\u4e00\u4e2a\u5408\u7406\u7684\u601d\u8def\u5c31\u662f\uff0c\u5728\u79bb\u6563\u7a7a\u95f4\u4e2d\u6784\u9020\u51fa\u76f8\u5e94\u7684Fokker-Planck Equation<\/span>\uff08\u4e5f\u53eb\u505aWasserstein\u68af\u5ea6\u6d41\uff09<\/span>\u548cLangevin Dynamics\uff0c\u4ece\u800c\u5f62\u6210\u53ef\u7528\u7684\u7b97\u6cd5\u3002\u76f8\u4fe1\u8fd9\u4e2a\u601d\u8def\u5bf9\u4e8e\u8bbe\u8ba1\u5176\u5b83\u79bb\u6563\u7a7a\u95f4\u7684\u7b97\u6cd5\u540c\u6837\u6709\u6240\u5e2e\u52a9\u3002<\/span><\/section>\n

      <\/span><\/section>\n
      \n
      \n
      \n
      Sampling as first-order optimization over a space of probability measures [2022]. Anna Korba, Adil Salim.<\/span><\/section>\n
      The variational formulation of the Fokker-Planck equation [1998]. Richard Jordan et al.<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n


      <\/p>\n

      <\/section>\n
      \u56fe2 \u5bf9Langevin Monte Carlo\u7b97\u6cd5\u7684\u7406\u89e3<\/span><\/section>\n

      <\/section>\n
      \u5c31\u5b9e\u73b0\u7ec6\u8282\u800c\u8a00\uff0c\u9996\u5148\u6ce8\u610f\u5230\u8fde\u7eed\u7a7a\u95f4\u548c\u79bb\u6563\u7a7a\u95f4\u7684\u6982\u7387\u51fd\u6570\u7a7a\u95f4\u5b9a\u4e49\u662f\u4e0d\u540c\u7684\u3002\u5bf9\u4e8e\u8fde\u7eed\u7a7a\u95f4\uff0c\u6982\u7387\u51fd\u6570\u53ef\u4ee5\u5b9a\u4e49\u4e3a\u6b27\u5f0f\u7a7a\u95f4\u4e2d\u7684\u4e00\u4e2a\u70b9\uff1b\u4f46\u5bf9\u4e8e\u79bb\u6563\u7a7a\u95f4\uff0c\u5219\u5e94\u5c06\u5b83\u5b9a\u4e49\u4e3a\u56fe<\/span>\uff08graph\uff09<\/span>\u4e0a\u7684\u4e00\u4e2a\u7ed3\u70b9\u3002\u4ece\u5df2\u6709\u6587\u732e\u7684\u7ed3\u8bba\uff0c\u53ef\u4ee5\u63a8\u5bfc\u51fa\u5728\u79bb\u6563\u7a7a\u95f4\u4e2dWasserstein\u68af\u5ea6\u6d41\u7684\u516c\u5f0f\u3002\u7136\u800c\u8fdb\u4e00\u6b65\u63a8\u5bfc\u79bb\u6563Langevin Dynamics\u5219\u662f\u56f0\u96be\u7684\uff0c\u4e3b\u8981\u539f\u56e0\u662f\u6b64\u65f6\u72b6\u6001\u7a7a\u95f4\u4e3a\u6307\u6570\u5927\u5c0f\uff0c\u4f8b\u5982\u6700\u5927\u5272\u95ee\u9898\u7684\u72b6\u6001\u7a7a\u95f4\u662f\u4e00\u4e2a\u8d85\u7acb\u65b9\u3001\u65c5\u884c\u5546\u95ee\u9898\u7684\u72b6\u6001\u7a7a\u95f4\u662f\u4e00\u4e2a\u5e8f\u5217\u7b49\u3002\u4e3a\u6b64\uff0c\u4f5c\u8005\u501f\u9274\u4e86\u5316\u5b66\u4e2d\u7535\u5bfc\u7387<\/span>\uff08conductance\uff09<\/span>\u7684\u6982\u5ff5\uff0c\u5b9a\u4e49\u4e86\u4e00\u79cd\u5408\u9002\u7684\u6743\u91cd\u516c\u5f0f\uff0c\u4ece\u800c\u63a8\u5bfc\u51fa\u79bb\u6563Langevin Dynamics\u7684\u5f62\u5f0f\uff1a<\/sub><\/span>\u3002\u8fdb\u800c\uff0c\u53ef\u4ee5\u901a\u8fc7\u6807\u51c6\u5316\u6a21\u62df\u65f6\u95f4\uff0c\u8ba1\u7b97\u8f6c\u79fb\u6982\u7387\u77e9\u9635\uff0c\u4f7f\u7528\u79bb\u6563\u9a6c\u5c14\u79d1\u592b\u94fe\u7684\u65b9\u6cd5\u8fdb\u884c\u91c7\u6837\uff1b\u4e5f\u53ef\u4ee5\u6807\u51c6\u5316\u6bcf\u4e00\u6b65\u7684\u6c49\u660e\u8ddd\u79bb\uff0c\u4f7f\u7528\u79f0\u4e3aPath Auxiliary Proposal\u7684\u65b9\u6cd5\u8fdb\u884c\u91c7\u6837\u3002\u56fe3\u5c55\u793a\u4e86\u6240\u63d0\u51fa\u7684\u68af\u5ea6\u4e0b\u964d\u91c7\u6837\u7b97\u6cd5\u4e0eGibbs\u8fd9\u79cd\u7b97\u6cd5\u76f8\u6bd4\u7684\u4f18\u52bf\u3002<\/span><\/section>\n

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

      Fokker-Planck equations for a free energy functional or Markov process on a graph [2012]. Shui-Nee Chow et al.<\/span><\/p>\n

      Discrete Langevin sampler via Wasserstein gradient Flow [2023]. Haoran Sun et al.<\/span><\/section>\n
      Thermodynamics of stoichiometric biochemical networks in living systems far from equilibrium [2005]. Hong Qian, Daniel A Beard.<\/span><\/section>\n
      Path auxiliary proposal for MCMC in discrete space [2022]. Haoran Sun et al.<\/span><\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n
      \n


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

      <\/section>\n
      \u56fe3 \u68af\u5ea6\u4e0b\u964d\u91c7\u6837\u4e0eGibbs\u91c7\u6837\u7684\u5bf9\u6bd4\u793a\u610f\u56fe<\/span><\/section>\n

      <\/section>\n

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

      <\/section>\n

      \n
      \n
      \n
      \n
      \n

      <\/section>\n

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

      3. \u91c7\u6837\u7b97\u6cd5\u4e0e\u7ec4\u5408\u4f18\u5316 <\/span><\/strong><\/strong><\/p>\n<\/section>\n

      \n

      <\/section>\n

      <\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
       <\/span><\/section>\n
      \u91c7\u6837\u7b97\u6cd5\u6700\u521d\u7684\u5e94\u7528\u662f\u5728\u5f88\u591a\u7269\u7406\u6a21\u578b\u4e2d\uff0c\u4f8b\u5982Ising\u6a21\u578b\u7b49\u3002\u672c\u5de5\u4f5c\u63d0\u51fa\u7684\u91c7\u6837\u65b9\u6cd5\u80fd\u591f\u5728\u8fd9\u4e9b\u975e\u5f52\u4e00\u5316\u5206\u5e03\u7684\u6a21\u578b\u4e0a\u53d1\u6325\u4f5c\u7528\u3002\u8fdb\u4e00\u6b65\u5730\uff0c\u5728\u7ec4\u5408\u4f18\u5316\u4e2d\u4e5f\u5b8c\u5168\u7528\u5f97\u4e0a\u8fd9\u79cd\u91c7\u6837\u6280\u672f\u3002\u8003\u8651\u4e00\u822c\u7684\u7ec4\u5408\u4f18\u5316\u95ee\u9898\uff0c\u5373min a(x), s.t. b(x)=0\uff0c\u5c06\u5176\u8f6c\u5316\u4e3a\u7f5a\u51fd\u6570\u5f62\u5f0fmin f(x)=a(x<\/span>)+\u03bb\u00b7b(x)\u3002\u5b9a\u4e49<\/span>\u6e29\u5ea6\u03c4\u4e0b\u7684\u6982\u7387\u5206<\/span>\u5e03<\/sub><\/span><\/span>\uff0c\u5219\u53ef\u4ee5\u5bf9\u6982\u7387\u7a7a\u95f4<\/sub> <\/span>(<\/span>\u03c40<\/sub><\/span>>\u03c4<\/span>1<\/sub><\/span><\/span>\u2026\u03c4<\/span>T<\/sub><\/span>\u2192<\/span>0)\u4f5c\u91c7\u6837\uff0c\u6765\u6c42\u89e3\u7ec4\u5408\u4f18\u5316\u95ee\u9898\u3002<\/span>\u524d\u9762\u63d0\u5230\u7684Path Auxil<\/span>iary\u91c7\u6837\u4e5f\u53ef\u4ee5\u7528\u4e8e\u7ec4\u5408\u4f18\u5316\u95ee\u9898\uff0c\u6bcf\u6b65\u6539\u53d8\u4e00\u4e2a\u72b6\u6001\uff0c\u4ece\u800c\u5728\u76ee\u6807\u72b6\u6001\u6539\u53d8\u591a\u4e2a\u7ed3\u70b9\u3002<\/span><\/section>\n

      <\/span><\/section>\n
      <\/section>\n
      \u56fe4 \u5728\u6700\u5927\u72ec\u7acb\u96c6\u95ee\u9898\u4e0a\u7684\u5b9e\u9a8c\u7ed3\u679c<\/span><\/section>\n

      <\/span><\/section>\n
      <\/section>\n
      \u56fe5 \u5728\u6700\u5927\u5272\u95ee\u9898\u4e0a\u7684\u5b9e\u9a8c\u7ed3\u679c<\/span><\/section>\n

      <\/span><\/section>\n
      <\/section>\n
      \u56fe6 \u5728\u65c5\u884c\u5546\u95ee\u9898\u4e0a\u7684\u5b9e\u9a8c\u7ed3\u679c<\/span><\/section>\n

      <\/span><\/section>\n
      \u5728\u6700\u5927\u72ec\u7acb\u96c6\u95ee\u9898<\/strong>\u4e0a\u7684\u5b9e\u9a8c\u7ed3\u679c<\/span>\uff08\u56fe4\uff09<\/span>\u663e\u793a\uff0c\u91c7\u6837\u7b97\u6cd5\u5728\u5f88\u591a\u5b9e\u4f8b\u4e0a\u627e\u5230\u4e86\u6bd4\u4f20\u7edf\u6c42\u89e3\u5668\u548c\u673a\u5668\u5b66\u4e60\u6c42\u89e3\u5668\u66f4\u4f18\u7684\u89e3\uff0c\u800c\u4e14\u8fd0\u884c\u65f6\u95f4\u4e5f\u76f8\u5f53\u6709\u4f18\u52bf\u3002\u800c\u8003\u8651\u5230\u91c7\u6837\u7b97\u6cd5\u7684\u8f7b\u91cf\u7ea7\u3001\u65e0\u9700\u8bad\u7ec3\u7684\u7279\u70b9\uff0c\u8fd9\u6837\u7684\u7ed3\u679c\u65e0\u7591\u662f\u975e\u5e38\u6709\u8da3\u7684\u3002\u5728\u6700\u5927\u5272\u95ee\u9898<\/strong>\u4e0a\u7684\u5b9e\u9a8c\u7ed3\u679c<\/span>\uff08\u56fe5\uff09<\/span>\u4e5f\u663e\u793a\uff0c\u5f53\u56fe\u4e2d\u7684\u7ed3\u70b9\u6570\u8f83\u591a\u65f6\uff0c\u91c7\u6837\u7b97\u6cd5\u5f97\u5230\u7684\u89e3\u5c06\u4f18\u4e8eGurobi\u3002\u5bf9\u4e8e\u65c5\u884c\u5546\u95ee\u9898<\/strong><\/span>\uff08\u56fe6\uff09<\/span>\uff0c\u91c7\u6837\u7b97\u6cd5\u867d\u7136\u4e0d\u53caLKH\uff0c\u4f46\u662f\u4e5f\u5f97\u5230\u4e86\u76f8\u5f53\u597d\u7684\u7ed3\u679c\uff0c\u5c24\u5176\u662f\u4e0e\u9700\u8981\u5927\u91cf\u8bad\u7ec3\u7684\u4e00\u7cfb\u5217\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u5bf9\u6bd4\u4e5f\u5177\u6709\u660e\u663e\u7684\u4f18\u52bf\u3002\u662f\u65f6\u5019\u5728\u7ec4\u5408\u4f18\u5316\u4e2d\u91cd\u65b0\u5ba1\u89c6\u91c7\u6837\u7b97\u6cd5\u4e86\u3002<\/span><\/section>\n

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

      <\/p>\n

      <\/span><\/section>\n
      \n
      \n
      Revisiting Sampling for Combinatorial Optimization [2023]. Haoran Sun et al.<\/span><\/section>\n<\/section>\n<\/blockquote>\n

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

      <\/section>\n

      \n
      \n
      \n
      \n
      \n

      <\/section>\n

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

      4. \u91c7\u6837\u7b97\u6cd5\u4e0eAI\u6a21\u578b<\/span><\/strong><\/strong><\/p>\n<\/section>\n

      \n

      <\/section>\n

      <\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
       <\/span><\/section>\n
      \u9664\u4e86\u6bd4\u8f83\u7ecf\u5178\u7684\u7ec4\u5408\u4f18\u5316\u95ee\u9898\u4e4b\u5916\uff0c\u5728\u56fe\u50cf\u548c\u6587\u672c\u751f\u6210\u7b49\u4efb\u52a1\u4e2d\uff0c\u91c7\u6837\u7b97\u6cd5\u4e5f\u80fd\u591f\u5f97\u5230\u4e00\u4e9b\u975e\u5e38\u6709\u8da3\u7684\u7ed3\u679c\u3002\u4f8b\u5982\uff0c\u53ef\u4ee5\u7ed9\u5b9a\u4e00\u4e2a\u80fd\u91cf\u6a21\u578b\uff0c\u5bf9\u56fe\u50cf\u8fdb\u884c\u91c7\u6837\u751f\u6210\uff1b\u6216\u662f\u7ed9\u53e5\u5b50\u53bb\u6389\u51e0\u4e2a\u8bcd\uff0c\u4f7f\u7528\u91c7\u6837\u7b97\u6cd5\u5bf9\u53bb\u6389\u7684\u8bcd\u505a\u51fa\u9884\u6d4b\u3002\u56fe7\u663e\u793a\uff0c\u597d\u7684\u91c7\u6837\u7b97\u6cd5\u53ef\u4ee5\u5f97\u5230\u66f4\u5177\u591a\u6837\u6027\u7684\u7ed3\u679c\u3002\u53e6\u4e00\u4e2a\u6700\u8fd1\u7684\u5c1d\u8bd5\u662f\u201c\u56fe\u751f\u6587\u201d\u4efb\u52a1\uff0c\u5373\u7ed9\u5b9a\u4e00\u5f20\u56fe\u50cf\uff0c\u9884\u6d4b\u5b83\u662f\u7528\u4ec0\u4e48\u63d0\u793a\u8bcd\u751f\u6210\u7684\u3002\u5982\u679c\u5c06\u8be5\u4efb\u52a1\u770b\u4f5c\u4e00\u79cd\u7ec4\u5408\u4f18\u5316\u95ee\u9898\uff0c\u5b83\u751a\u81f3\u4e0d\u80fd\u4e25\u683c\u5730\u8ba1\u7b97\u51fa\u76ee\u6807\u51fd\u6570\u3002\u5982\u56fe8\u6240\u793a\uff0c\u4f5c\u8005\u63d0\u51fa\u7684\u4e00\u79cd\u57fa\u4e8e\u68af\u5ea6\u4e0b\u964d\u7684\u91c7\u6837\u7b97\u6cd5\u540c\u6837\u53ef\u4ee5\u5728\u8be5\u4efb\u52a1\u4e2d\u8868\u73b0\u826f\u597d\u3002\u8fd9\u4e9b\u5e94\u7528\u90fd\u4f53\u73b0\u51fa\u91c7\u6837\u7b97\u6cd5\u4e0d\u4ec5\u9002\u7528\u4e8e\u7ecf\u5178\u7684\u7ec4\u5408\u4f18\u5316\uff0c\u8fd8\u80fd\u591f\u5728\u5f88\u591a\u4e0e\u57fa\u7840\u6a21\u578b\u6709\u5173\u7684\u4efb\u52a1\u4e2d\u53d1\u6325\u4f5c\u7528\u3002<\/span><\/section>\n

      <\/span><\/section>\n
      <\/section>\n
      \u56fe7 \u91c7\u6837\u7b97\u6cd5\u5728\u56fe\u50cf\u548c\u6587\u672c\u751f\u6210\u4e2d\u7684\u5e94\u7528<\/span><\/section>\n

      <\/span><\/section>\n
      <\/section>\n
      \u56fe8 \u91c7\u6837\u7b97\u6cd5\u5728\u56fe\u201c\u56fe\u751f\u6587\u201d\u4efb\u52a1\u4e2d\u7684\u5e94\u7528<\/span><\/section>\n

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

      <\/section>\n

      <\/section>\n
      \u626b\u63cf\u4e0b\u65b9\u4e8c\u7ef4\u7801\u89c2\u770b\u8bfb\u4e66\u4f1a\u89c6\u9891\u56de\u653e<\/span><\/section>\n
      <\/section>\n

      <\/section>\n

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

      \u5b66\u8005\u7b80\u4ecb<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n

      <\/p>\n

      \u6234\u6db5\u4fca<\/span><\/strong>\uff0cGoogle DeepMind\u5b9e\u9a8c\u5ba4\u8d44\u6df1\u79d1\u5b66\u5bb6\u548c\u7814\u7a76\u7ecf\u7406\uff0c\u6b64\u524d\u4e8e\u4f50\u6cbb\u4e9a\u7406\u5de5\u83b7\u5f97\u535a\u58eb\u5b66\u4f4d\u3002<\/span>\u7814\u7a76\u65b9\u5411\uff1a\u9ad8\u6548\u7684\u751f\u6210\u6a21\u578b\uff0c\u5305\u62ec\u5927\u8bed\u8a00\u6a21\u578b\uff0c\u56fe\u50cf\u548c\u7ed3\u6784\u5316\u6570\u636e\u6a21\u578b\uff0c\u53ca\u5176\u91c7\u6837\u548c\u4f18\u5316\u7684\u5e95\u5c42\u7b97\u6cd5<\/span><\/h4>\n


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


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

      \u56fe\u795e\u7ecf\u7f51\u7edc\u4e0e\u7ec4\u5408\u4f18\u5316\u8bfb\u4e66\u4f1a<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

      \n


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

      \u73b0\u5b9e\u4e16\u754c\u4e2d\u5927\u91cf\u95ee\u9898\u7684\u89e3\u51b3\u4f9d\u8d56\u4e8e\u7b97\u6cd5\u7684\u8bbe\u8ba1\u4e0e\u6c42\u89e3\u3002\u4f20\u7edf\u7b97\u6cd5\u7531\u4eba\u7c7b\u4e13\u5bb6\u8bbe\u8ba1\uff0c\u800c\u968f\u7740\u4eba\u5de5\u667a\u80fd\u6280\u672f\u4e0d\u65ad\u53d1\u5c55\uff0c\u7b97\u6cd5\u81ea\u52a8\u5b66\u4e60\u7b97\u6cd5\u7684\u6848\u4f8b\u65e5\u76ca\u589e\u591a\uff0c\u5982\u4ee5\u795e\u7ecf\u7f51\u7edc\u4e3a\u4ee3\u8868\u7684\u7684\u4eba\u5de5\u667a\u80fd\u7b97\u6cd5\uff0c\u8fd9\u662f\u7b97\u6cd5\u795e\u7ecf\u5316\u6c42\u89e3\u7684\u7f18\u7531\u3002\u5728\u7b97\u6cd5\u795e\u7ecf\u5316\u6c42\u89e3\u65b9\u5411\u4e0a\uff0c\u56fe\u795e\u7ecf\u7f51\u7edc\u662f\u4e00\u4e2a\u5f3a\u6709\u529b\u7684\u5de5\u5177\uff0c\u80fd\u591f\u5145\u5206\u5229\u7528\u56fe\u7ed3\u6784\u7684\u7279\u6027\uff0c\u5b9e\u73b0\u5bf9\u9ad8\u590d\u6742\u5ea6\u7b97\u6cd5\u7684\u9ad8\u6548\u8fd1\u4f3c\u6c42\u89e3\u3002\u57fa\u4e8e\u56fe\u795e\u7ecf\u7f51\u7edc\u7684\u590d\u6742\u7cfb\u7edf\u4f18\u5316\u4e0e\u63a7\u5236\u5c06\u4f1a\u662f\u5927\u6a21\u578b\u70ed\u6f6e\u4e4b\u540e\u65b0\u7684\u672a\u6765\u65b9\u5411\u3002<\/span><\/section>\n

      <\/span><\/section>\n
      \u4e3a\u4e86\u63a2\u8ba8\u56fe\u795e\u7ecf\u7f51\u7edc\u5728\u7b97\u6cd5\u795e\u7ecf\u5316\u6c42\u89e3\u7684\u53d1\u5c55\u4e0e\u73b0\u5b9e\u5e94\u7528\uff0c\u96c6\u667a\u4ff1\u4e50\u90e8\u8054\u5408\u56fd\u9632\u79d1\u6280\u5927\u5b66\u7cfb\u7edf\u5de5\u7a0b\u5b66\u9662\u526f\u6559\u6388\u8303\u957f\u4fca\u3001\u4e2d\u56fd\u4eba\u6c11\u5927\u5b66\u9ad8\u74f4\u4eba\u5de5\u667a\u80fd\u5b66\u9662\u526f\u6559\u6388\u9ec4\u6587\u70b3\uff0c\u5171\u540c\u53d1\u8d77\u300c\u56fe\u795e\u7ecf\u7f51\u7edc\u4e0e\u7ec4\u5408\u4f18\u5316\u300d\u8bfb\u4e66\u4f1a\u3002\u8bfb\u4e66\u4f1a\u5c06\u805a\u7126\u4e8e\u56fe\u795e\u7ecf\u7f51\u7edc\u4e0e\u7b97\u6cd5\u795e\u7ecf\u5316\u6c42\u89e3\u7684\u76f8\u5173\u9886\u57df\uff0c\u5305\u62ec\u795e\u7ecf\u7b97\u6cd5\u63a8\u7406\u3001\u7ec4\u5408\u4f18\u5316\u95ee\u9898\u6c42\u89e3\u3001\u51e0\u4f55\u56fe\u795e\u7ecf\u7f51\u7edc\uff0c\u4ee5\u53ca\u7b97\u6cd5\u795e\u7ecf\u5316\u6c42\u89e3\u5728 AI for Science \u4e2d\u7684\u5e94\u7528\u7b49\u65b9\u9762\uff0c\u5e0c\u671b\u4e3a\u53c2\u4e0e\u8005\u63d0\u4f9b\u4e00\u4e2a\u5b66\u672f\u4ea4\u6d41\u5e73\u53f0\uff0c\u6fc0\u53d1\u53c2\u4e0e\u8005\u7684\u5b66\u672f\u5174\u8da3\uff0c\u8fdb\u4e00\u6b65\u63a8\u52a8\u76f8\u5173\u9886\u57df\u7684\u7814\u7a76\u548c\u5e94\u7528\u53d1\u5c55\u3002\u6b22\u8fce\u611f\u5174\u8da3\u7684\u670b\u53cb\u62a5\u540d\u53c2\u4e0e\uff01<\/span><\/section>\n

      <\/span><\/section>\n
      <\/span><\/a><\/section>\n
      <\/span><\/section>\n
      \u8be6\u60c5\u8bf7\u89c1\uff1a<\/span><\/span><\/section>\n
      \u52a0\u901f\u7ecf\u5178\u7b97\u6cd5\u6548\u7387\uff0c\u7a81\u7834\u73b0\u5b9e\u6280\u672f\u74f6\u9888\uff1a\u56fe\u795e\u7ecf\u7f51\u7edc\u4e0e\u7ec4\u5408\u4f18\u5316\u8bfb\u4e66\u4f1a\u542f\u52a8<\/a><\/span><\/section>\n

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

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

      \u63a8\u8350\u9605\u8bfb<\/span><\/strong><\/strong><\/p>\n

      1. <\/strong>\u591a\u6a21\u6001\u56fe\u5b66\u4e60\u84dd\u56fe\uff1a\u5e94\u7528\u4e8e\u89c6\u89c9\u3001\u8bed\u8a00\u3001\u81ea\u7136\u79d1\u5b66<\/a><\/strong><\/span><\/strong><\/section>\n
      2.<\/strong> \u4eba\u5de5\u667a\u80fd\u524d\u6cbf\uff1a\u7ec4\u5408\u4f18\u5316\u95ee\u9898\u7684\u673a\u5668\u5b66\u4e60\u6c42\u89e3<\/a> <\/strong><\/span><\/strong><\/section>\n
      3.<\/strong> \u51e0\u4f55\u6df1\u5ea6\u5b66\u4e60\uff1a\u8ba9\u7269\u7406\u4e16\u754c\u62e5\u6709AI<\/a><\/strong><\/strong><\/span><\/section>\n
      4.<\/strong> \u56fe\u4e0e\u57fa\u7840\u6a21\u578b\uff1a\u591a\u6a21\u6001\u57fa\u7840\u6a21\u578b\u5173\u7cfb\u63a8\u7406\u80fd\u529b\u6982\u8ff0<\/a><\/strong><\/strong><\/span><\/strong><\/strong><\/span><\/section>\n
      5. <\/strong><\/span>\u5f20\u6c5f\uff1a\u7b2c\u4e09\u4ee3\u4eba\u5de5\u667a\u80fd\u6280\u672f\u57fa\u7840\u2014\u2014\u4ece\u53ef\u5fae\u5206\u7f16\u7a0b\u5230\u56e0\u679c\u63a8\u7406 | \u96c6\u667a\u5b66\u56ed\u5168\u65b0\u8bfe\u7a0b<\/a><\/strong><\/strong><\/span><\/section>\n
      6. \u6210\u4e3a\u96c6\u667aVIP\uff0c\u89e3\u9501\u5168\u7ad9\u8bfe\u7a0b\uff0f\u8bfb\u4e66\u4f1a<\/a><\/span><\/strong><\/strong><\/strong><\/strong><\/strong><\/span><\/section>\n
      7. <\/strong>\u52a0\u5165\u96c6\u667a\uff0c\u4e00\u8d77\u590d\u6742\uff01<\/strong><\/span><\/a><\/section>\n


      <\/p>\n


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

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

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

      \u5bfc\u8bed \u5728\u96c6\u667a\u4ff1\u4e50\u90e8\u300c\u56fe\u795e\u7ecf\u7f51\u7edc\u4e0e\u7ec4\u5408\u4f18\u5316\u8bfb\u4e66\u4f1a\u300d\u7b2c\u4e00\u5b63\u6700\u540e\u4e00\u671f\uff0cDeepMind \u5b9e\u9a8c\u5ba4\u8d44\u6df1\u79d1\u5b66\u5bb6\u6234\u6db5\u4fca\u9996\u5148\u4ecb\u7ecd\u4e86\u4ed6\u4eec\u8fd1\u671f\u53d1\u8868\u4e8e PMLR \u7684\u4e00\u9879\u6700\u65b0\u7814\u7a76\uff0c\u201c\u5728\u7ec4\u5408\u4f18\u5316\u95ee\u9898\u4e2d\u91cd\u65b0\u5ba1\u89c6\u62bd\u6837\u201d\uff08Revisiting Sampling for Combinatorial Optimization\uff09\u3002 \u7814\u7a76…<\/p>\n","protected":false},"author":0,"featured_media":46761,"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\/46779"}],"collection":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=46779"}],"version-history":[{"count":0,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/46779\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/media\/46761"}],"wp:attachment":[{"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=46779"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=46779"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=46779"},{"taxonomy":"special","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fspecial&post=46779"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}