{"id":59053,"date":"2025-04-28T22:14:55","date_gmt":"2025-04-28T14:14:55","guid":{"rendered":"https:\/\/swarma.org\/?p=59053"},"modified":"2025-04-28T22:14:55","modified_gmt":"2025-04-28T14:14:55","slug":"%e3%80%8aaixscience%e5%8d%81%e5%a4%a7%e5%89%8d%e6%b2%bf%e8%a7%82%e5%af%9f%e3%80%8b2%ef%bc%9a%e5%9e%82%e7%9b%b4%e9%a2%86%e5%9f%9f%e7%a7%91%e5%ad%a6%e5%a4%a7%e6%a8%a1%e5%9e%8b","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=59053","title":{"rendered":"\u300aAI\u00d7SCIENCE\u5341\u5927\u524d\u6cbf\u89c2\u5bdf\u300b2\uff1a\u5782\u76f4\u9886\u57df\u79d1\u5b66\u5927\u6a21\u578b"},"content":{"rendered":"
\n
\n
\n
\n
\n
\n
\n
<\/span><\/section>\n
\n
\n
\n
\n
\n

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

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

\n

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

\u4e0a\u6d77\u79d1\u5b66\u667a\u80fd\u7814\u7a76\u9662\u3001\u96c6\u667a\u79d1\u5b66\u7814\u7a76\u4e2d\u5fc3\u548c\u963f\u91cc\u4e91\u8054\u5408\u53d1\u5e03\u4e86\u300aAI X Science\u5341\u5927\u524d\u6cbf\u89c2\u5bdf\u300b\uff0c\u68b3\u7406\u51fa35\u4e2a\u7814\u7a76\u524d\u6cbf\uff0c\u6765\u63a8\u52a8\u79d1\u5b66\u53d1\u5c55\u7684\u9ec4\u91d1\u65f6\u4ee3\u5230\u6765\u3002\u672c\u7bc7\u4e3a\u524d\u6cbf\u89c2\u5bdf2\uff0c\u626b\u63cf\u4e0b\u65b9\u4e8c\u7ef4\u7801\uff0c\u53ef\u83b7\u5f97\u5b8c\u6574\u7248\u4e0b\u8f7d\u5730\u5740\uff0c\u5e76\u80fd\u5feb\u901f\u94fe\u63a5\u8bba\u6587\u539f\u6587\u3002<\/span><\/span><\/strong><\/p>\n


<\/span><\/strong><\/section>\n
<\/span><\/strong><\/section>\n
\u626b\u63cf\u4e8c\u7ef4\u7801 \u4e0b\u8f7d\u5b8c\u6574\u62a5\u544a<\/span><\/strong><\/section>\n<\/section>\n
<\/strong><\/strong><\/section>\n<\/section>\n
\n
<\/strong><\/strong><\/section>\n<\/section>\n
\n
<\/strong><\/strong><\/section>\n
<\/strong><\/strong><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\n
\n
\n
<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\n
\"\u56fe\u7247\"<\/section>\n<\/section>\n<\/section>\n<\/section>\n


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

\n
\n
\n
\n
\n

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

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

\u5782\u76f4\u9886\u57df\u79d1\u5b66\u5927\u6a21\u578b<\/span><\/strong><\/span><\/strong><\/strong><\/p>\n<\/section>\n

\n

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

<\/span><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n


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

\u80cc\u666f\u4ecb\u7ecd<\/span><\/span><\/p>\n

<\/section>\n

\u4eba\u5de5\u667a\u80fd\u7684\u8303\u5f0f\u8f6c\u53d8<\/span><\/span><\/em><\/span><\/p>\n

\u8fd1\u5e74\u6765\uff0c\u4ee5GPT\u548cBERT\u4e3a\u4ee3\u8868\u7684\u57fa\u7840\u6a21\u578b<\/span> (Foundation Models)<\/span> [1]\u5728\u4eba\u5de5\u667a\u80fd\u9886\u57df\u53d6\u5f97\u4e86\u7a81\u7834\u6027\u8fdb\u5c55\u3002\u8fd9\u7c7b\u6a21\u578b\u901a\u8fc7\u5927\u89c4\u6a21\u9884\u8bad\u7ec3\u83b7\u5f97\u4e86\u5f3a\u5927\u7684\u8868\u793a\u5b66\u4e60\u80fd\u529b\u548c\u826f\u597d\u7684\u8fc1\u79fb\u80fd\u529b\uff0c\u4e3a\u4eba\u5de5\u667a\u80fd\u7684\u53d1\u5c55\u5f00\u8f9f\u4e86\u65b0\u65b9\u5411\u3002\u7136\u800c\uff0c\u5728\u4e13\u4e1a\u79d1\u5b66\u9886\u57df\u7684\u5e94\u7528\u4e2d\uff0c\u7814\u7a76\u8005\u4eec\u53d1\u73b0\u901a\u7528\u57fa\u7840\u6a21\u578b\u5f80\u5f80\u96be\u4ee5\u76f4\u63a5\u5e94\u7528\u4e8e\u7279\u5b9a\u79d1\u5b66\u95ee\u9898\u7684\u89e3\u51b3\u3002\u8fd9\u4e3b\u8981\u6e90\u4e8e\u79d1\u5b66\u7814\u7a76\u5bf9\u4e13\u4e1a\u77e5\u8bc6\u6df1\u5ea6\u6574\u5408\u7684\u9700\u6c42\u3001\u6570\u636e\u7279\u70b9\u7684\u5dee\u5f02\u6027\u4ee5\u53ca\u4efb\u52a1\u7279\u5f81\u7684\u7279\u6b8a\u6027\u3002<\/span><\/span><\/p>\n

\u5728\u79d1\u5b66\u7814\u7a76\u9886\u57df\uff0c\u6a21\u578b\u4e0d\u4ec5\u9700\u8981\u5177\u5907\u5f3a\u5927\u7684\u5b66\u4e60\u80fd\u529b\uff0c\u66f4\u9700\u8981\u4e25\u683c\u9075\u5faa\u7269\u7406\u89c4\u5f8b\uff0c\u4fdd\u6301\u7ed3\u679c\u7684\u53ef\u89e3\u91ca\u6027\uff0c\u5e76\u786e\u4fdd\u9884\u6d4b\u7684\u7269\u7406\u5408\u7406\u6027\u3002\u4f8b\u5982\uff0c\u5728\u5206\u5b50\u8bbe\u8ba1\u4e2d\uff0c\u6a21\u578b\u9700\u8981\u786e\u4fdd\u9884\u6d4b\u7ed3\u6784\u6ee1\u8db3\u5316\u5b66\u952e\u89d2\u548c\u952e\u957f\u7684\u7ea6\u675f\uff1b\u5728\u6750\u6599\u7814\u7a76\u4e2d\uff0c\u9700\u8981\u8003\u8651\u6676\u4f53\u7ed3\u6784\u7684\u5468\u671f\u6027\u548c\u5bf9\u79f0\u6027\u3002\u8fd9\u4e9b\u7279\u6b8a\u9700\u6c42\u63a8\u52a8\u4e86\u9488\u5bf9\u5355\u4e00\u9886\u57df\u7684\u57fa\u7840\u6a21\u578b\u7814\u7a76\u3002\u4ee5AlphaFold3[2]\u4e3a\u4ee3\u8868\u7684\u6210\u529f\u6848\u4f8b\uff0c\u5c55\u793a\u4e86\u5c06\u57fa\u7840\u6a21\u578b\u601d\u60f3\u4e0e\u9886\u57df\u4e13\u4e1a\u77e5\u8bc6\u6df1\u5ea6\u878d\u5408\u7684\u5de8\u5927\u6f5c\u529b\u3002<\/span><\/span><\/p>\n

\u5355\u4e00\u9886\u57df\u57fa\u7840\u6a21\u578b\u7ee7\u627f\u4e86\u57fa\u7840\u6a21\u578b\u7684\u6838\u5fc3\u7279\u5f81\uff0c\u5982\u5927\u89c4\u6a21\u9884\u8bad\u7ec3\u3001\u901a\u7528\u8868\u793a\u5b66\u4e60\u548c\u8fc1\u79fb\u80fd\u529b\uff0c\u540c\u65f6\u6df1\u5ea6\u6574\u5408\u4e86\u9886\u57df\u7279\u5b9a\u77e5\u8bc6\uff0c\u5305\u62ec\u7269\u7406\u89c4\u5f8b\u3001\u4e13\u4e1a\u77e5\u8bc6\u548c\u9886\u57df\u7ea6\u675f\u3002\u4e0e\u901a\u7528\u57fa\u7840\u6a21\u578b\u76f8\u6bd4\uff0c\u9886\u57df\u57fa\u7840\u6a21\u578b\u5728\u6570\u636e\u7279\u70b9\u3001\u4efb\u52a1\u7279\u5f81\u548c\u5e94\u7528\u573a\u666f\u7b49\u65b9\u9762\u90fd\u5448\u73b0\u51fa\u663e\u8457\u7684\u7279\u5f02\u6027\u3002<\/span><\/span><\/p>\n

\u4ece\u53d1\u5c55\u5386\u7a0b\u6765\u770b\uff0c\u9886\u57df\u57fa\u7840\u6a21\u578b\u7684\u6f14\u8fdb\u5927\u81f4\u53ef\u5206\u4e3a\u4e09\u4e2a\u9636\u6bb5\u3002\u5728\u65e9\u671f\u63a2\u7d22\u9636\u6bb5(2017-2019)\uff0c\u7814\u7a76\u4e3b\u8981\u96c6\u4e2d\u5728\u7279\u5b9a\u4efb\u52a1\u7684\u6df1\u5ea6\u5b66\u4e60\u65b9\u6cd5\u4e0a\uff0c\u5982DeepChem[3]\u5728\u5206\u5b50\u6027\u8d28\u9884\u6d4b\u4e2d\u7684\u5e94\u7528\uff0c\u4ee5\u53ca\u57fa\u4e8e3D-CNN[4]\u7684\u6750\u6599\u7ed3\u6784\u5206\u6790\uff1b\u5feb\u901f\u53d1\u5c55\u9636\u6bb5\uff082020-2021\uff09\u89c1\u8bc1\u4e86\u51e0\u4e2a\u5177\u6709\u91cc\u7a0b\u7891\u610f\u4e49\u7684\u7a81\u7834\uff0c\u5176\u4e2d\u6700\u5177\u4ee3\u8868\u6027\u7684\u662fAlphaFold1[5]\u5728\u86cb\u767d\u8d28\u7ed3\u6784\u9884\u6d4b\u9886\u57df\u7684\u91cd\u5927\u7a81\u7834\u3002\u8fdb\u5165\u6210\u719f\u5e94\u7528\u9636\u6bb5\uff082022-2023\uff09\uff0c\u6211\u4eec\u770b\u5230\u4e86\u66f4\u591a\u521b\u65b0\u6027\u7684\u5de5\u4f5c\uff1a\u7269\u8d28\u79d1\u5b66\u9886\u57df\u4e2d\u9488\u5bf9\u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df\u7684GPTFF<\/span>\uff08Graph-based Pre-trained Transformer Force Field\uff09<\/span>[6]\u4ee5\u53ca\u9762\u5411\u9ad8\u6548\u6269\u5c55\u795e\u7ecf\u7b97\u5b50\u7684\u7edf\u4e00\u6846\u67b6Universal Physics Transformer (UPT)[7]\uff0c\u8fd8\u6709\u751f\u547d\u79d1\u5b66\u9886\u57df\u7684scFoundation[8]\u548cGMAI[9]\u548c\u5730\u7403\u6c14\u5019\u7cfb\u7edf\u9886\u57df\u7684Pangu-Weather[10]\u3001FuXi[11]\u3001FengWu[12]\u7b49\u5de5\u4f5c\uff0c\u90fd\u8d8a\u6765\u8d8a\u8d70\u5411\u79d1\u5b66\u7edf\u4e00\u7684\u57fa\u7840\u6a21\u578b\u3002<\/span><\/span><\/p>\n

\u8fd9\u79cd\u53d1\u5c55\u8fc7\u7a0b\u59cb\u7ec8\u90fd\u662f\u56f4\u7ed5\u7740\u6838\u5fc3\u5173\u952e\u7684\u95ee\u9898\uff1a\u80fd\u5426\u5c06\u57fa\u7840\u6a21\u578b\u7684\u7406\u5ff5\u63a8\u5e7f\u5230\u66f4\u5e7f\u6cdb\u7684\u79d1\u5b66\u7814\u7a76\u9886\u57df\uff1f\u7279\u522b\u662f\uff0c\u80fd\u5426\u6784\u5efa\u4e00\u4e2a\u7edf\u4e00\u7684\u79d1\u5b66\u57fa\u7840\u6a21\u578b\uff0c\u6765\u52a0\u901f\u8de8\u5b66\u79d1\u7684\u79d1\u5b66\u53d1\u73b0\u8fdb\u7a0b\uff1f\u8981\u56de\u7b54\u8fd9\u4e2a\u95ee\u9898\uff0c\u4ecd\u9762\u4e34\u4e09\u4e2a\u6839\u672c\u6027\u7684\u6311\u6218[13]\uff1a<\/span><\/span><\/p>\n

1. \u5982\u4f55\u5904\u7406\u8de8\u6a21\u6001\u3001\u8de8\u5c3a\u5ea6\u3001\u8de8\u9886\u57df\u7684\u79d1\u5b66\u6570\u636e\uff1f<\/span><\/span><\/p>\n

2. \u5982\u4f55\u5c06\u8fd9\u4e9b\u57fa\u672c\u79d1\u5b66\u89c4\u5f8b\u878d\u5165\u6a21\u578b\u8bbe\u8ba1\uff0c\u4f7f\u6a21\u578b\u65e2\u4fdd\u6301\u7075\u6d3b\u6027\u53c8\u4e0d\u8fdd\u80cc\u57fa\u672c\u7269\u7406\u5b9a\u5f8b\uff1f<\/span><\/span><\/p>\n

3. \u5982\u4f55\u8bbe\u8ba1\u6a21\u578b\u67b6\u6784\uff0c\u4f7f\u5176\u80fd\u591f\u6355\u6349\u5e76\u5229\u7528\u8fd9\u4e9b\u6f5c\u5728\u5173\u8054\uff0c\u5b9e\u73b0\u8de8\u5b66\u79d1\u7684\u77e5\u8bc6\u8fc1\u79fb\u548c\u521b\u65b0\u53d1\u73b0\uff1f<\/span><\/span><\/p>\n

\u867d\u7136\u76ee\u524d\u8fd8\u6ca1\u6709\u51fa\u73b0\u7edf\u4e00\u7684\u79d1\u5b66\u57fa\u7840\u6a21\u578b\uff0c\u4f46\u662f\u7269\u8d28\u79d1\u5b66\u3001\u751f\u547d\u79d1\u5b66\u3001\u6c14\u8c61\u9884\u6d4b\u7b49\u5404\u4e2a\u9886\u57df\u5df2\u7ecf\u6d8c\u73b0\u51fa\u4e00\u4e9b\u521b\u65b0\u6027\u7684\u5de5\u4f5c\uff0c\u51fa\u73b0\u4e86\u5927\u91cf\u7684\u5355\u4e00\u9886\u57df\u79d1\u5b66\u57fa\u7840\u6a21\u578b\uff0c\u89e3\u51b3\u9886\u57df\u95ee\u9898\u3002\u4ee5\u4e0b\u6211\u4eec\u5c06\u91cd\u70b9\u4ecb\u7ecd\u4e09\u4e2a\u6700\u5177\u4ee3\u8868\u6027\u7684\u7814\u7a76\u65b9\u5411\u53ca\u5176\u6700\u65b0\u8fdb\u5c55\u3002<\/span><\/span><\/p>\n

\u7814\u7a76\u8fdb\u5c55<\/span><\/span><\/h2>\n
\n

\u8fdb\u5c55\u76ee\u5f55<\/span><\/span><\/p>\n

1. \u751f\u547d\u79d1\u5b66\u57fa\u7840\u6a21\u578b<\/span><\/span><\/p>\n

2. \u533b\u5b66\u57fa\u7840\u6a21\u578b<\/span><\/span><\/p>\n

3. \u6c14\u8c61\u9884\u6d4b\u57fa\u7840\u6a21\u578b<\/span><\/span><\/p>\n

4. \u7269\u8d28\u79d1\u5b66\u57fa\u7840\u6a21\u578b<\/span><\/span><\/p>\n

a. \u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df<\/span><\/span><\/p>\n

b. \u7269\u7406\u6a21\u62df<\/span><\/span><\/p>\n<\/blockquote>\n


<\/span><\/strong><\/h3>\n

\u751f\u547d\u79d1\u5b66\u57fa\u7840\u6a21\u578b<\/span><\/span><\/strong><\/h3>\n

\u63a8\u8350\u7406\u7531\uff1a<\/span><\/span><\/strong> scFoundation[8]\u4ee3\u8868\u4e86\u751f\u547d\u79d1\u5b66\u9886\u57df\u57fa\u7840\u6a21\u578b\u7684\u91cd\u8981\u7a81\u7834\u3002\u5b83\u5b9e\u73b0\u5bf9\u590d\u6742\u751f\u547d\u7cfb\u7edf\u7684\u5efa\u6a21\uff0c\u4e0d\u4ec5\u5728\u6280\u672f\u4e0a\u521b\u65b0\u6027\u5730\u89e3\u51b3\u4e86\u5355\u7ec6\u80de\u6d4b\u5e8f\u6570\u636e\u7684\u7279\u6b8a\u6311\u6218\uff0c\u66f4\u4e3a\u751f\u547d\u79d1\u5b66\u7684\u667a\u80fd\u5316\u5206\u6790\u5f00\u8f9f\u4e86\u65b0\u8303\u5f0f\u3002<\/span><\/span><\/p>\n

\u968f\u7740\u5355\u7ec6\u80deRNA\u6d4b\u5e8f\u6280\u672f\u7684\u53d1\u5c55\uff0c\u79d1\u7814\u4eba\u5458\u80fd\u591f\u5f00\u59cb\u89c2\u5bdf\u548c\u7406\u89e3\u7ec6\u80de\u5185\u7684\u57fa\u56e0\u8868\u8fbe\u6a21\u5f0f\u3002\u7136\u800c\uff0c\u5982\u4f55\u4ece\u6d77\u91cf\u3001\u5f02\u8d28\u3001\u5b58\u5728\u6280\u672f\u566a\u58f0\u7684\u5355\u7ec6\u80de\u6570\u636e\u4e2d\u5b66\u4e60\u5230\u6709\u610f\u4e49\u7684\u751f\u7269\u5b66\u77e5\u8bc6\uff0c\u4e00\u76f4\u662f\u8be5\u9886\u57df\u7684\u6838\u5fc3\u6311\u6218\u3002\u4f20\u7edf\u7684\u8ba1\u7b97\u65b9\u6cd5\u5f80\u5f80\u53ea\u80fd\u5904\u7406\u90e8\u5206\u57fa\u56e0\uff0c\u6216\u8005\u9700\u8981\u9488\u5bf9\u7279\u5b9a\u4efb\u52a1\u5355\u72ec\u6784\u5efa\u6a21\u578b\uff0c\u7f3a\u4e4f\u7edf\u4e00\u7684\u5206\u6790\u6846\u67b6\u3002<\/span><\/span><\/p>\n

\u57282023\u5e74Geneformer[14]\u9996\u6b21\u9a8c\u8bc1\u57fa\u56e0\u8868\u8fbe\u6570\u636e\u9884\u8bad\u7ec3\u7684\u53ef\u884c\u6027\uff0c\u4ee5\u53cascGPT[15]\u63a2\u7d22\u751f\u6210\u5f0f\u65b9\u6cd5\u7684\u57fa\u7840\u4e0a\uff0cscFoundation[8]\u901a\u8fc7\u72ec\u7279\u7684\u6280\u672f\u65b9\u6848\u5b9e\u73b0\u4e86\u5bf9\u590d\u6742\u751f\u547d\u7cfb\u7edf\u7684\u7edf\u4e00\u5efa\u6a21\uff0c\u6807\u5fd7\u7740\u751f\u547d\u79d1\u5b66\u6b63\u5728\u8fdb\u5165\u57fa\u7840\u6a21\u578b\u65f6\u4ee3\u3002<\/span><\/span><\/p>\n

<\/section>\n

scFoundation\u9884\u8bad\u7ec3\u6846\u67b6\u6982\u89c8 | \u6765\u6e90\uff1aHao, M., Gong, J., Zeng, X. et al. Large-scale foundation model on single-cell transcriptomics. Nat Methods 21, 1481\u20131491 (2024). https:\/\/doi.org\/10.1038\/s41592-024-02305-7<\/span><\/span><\/em><\/span><\/p>\n

scFoundation\u901a\u8fc7\u521b\u65b0\u7684\u6a21\u578b\u67b6\u6784\u8bbe\u8ba1\u548c\u9884\u8bad\u7ec3\u7b56\u7565\uff0c\u9996\u6b21\u5b9e\u73b0\u4e86\u5bf9\u5168\u57fa\u56e0\u7ec4\u8303\u56f4\u7684\u7edf\u4e00\u5efa\u6a21\u3002\u8be5\u6a21\u578b\u5305\u542b1\u4ebf\u4e2a\u53c2\u6570\uff0c\u8986\u76d6\u7ea62\u4e07\u4e2a\u86cb\u767d\u7f16\u7801\u57fa\u56e0\uff0c\u5728\u8d85\u8fc75000\u4e07\u4e2a\u4eba\u7c7b\u5355\u7ec6\u80de\u8f6c\u5f55\u7ec4\u6570\u636e\u4e0a\u8fdb\u884c\u9884\u8bad\u7ec3\u3002\u5176\u6838\u5fc3\u521b\u65b0\u5728\u4e8e\uff1a\u9996\u5148\uff0c\u8bbe\u8ba1\u4e86\u4e00\u4e2a\u975e\u5bf9\u79f0\u7684Transformer\u67b6\u6784<\/span> (xTrimoGene)<\/span>\uff0c\u901a\u8fc7\u5c06\u7f16\u7801\u5668\u9650\u5236\u5728\u975e\u96f6\u8868\u8fbe\u57fa\u56e0\u4e0a\u5904\u7406\u540c\u65f6\u8ba9\u89e3\u7801\u5668\u8986\u76d6\u5168\u57fa\u56e0\u7ec4\u8303\u56f4\uff0c\u5de7\u5999\u89e3\u51b3\u4e86\u6570\u636e\u7a00\u758f\u6027\u5e26\u6765\u7684\u8ba1\u7b97\u6311\u6218\uff1b\u5176\u6b21\uff0c\u63d0\u51fa\u4e86\u8bfb\u53d6\u6df1\u5ea6\u611f\u77e5<\/span> (RDA) <\/span>\u7684\u9884\u8bad\u7ec3\u4efb\u52a1\uff0c\u4e0d\u4ec5\u53ef\u4ee5\u6355\u83b7\u57fa\u56e0\u95f4\u7684\u590d\u6742\u5171\u8868\u8fbe\u5173\u7cfb\uff0c\u8fd8\u80fd\u81ea\u52a8\u5904\u7406\u4e0d\u540c\u6d4b\u5e8f\u6df1\u5ea6\u5e26\u6765\u7684\u6280\u672f\u566a\u58f0\uff0c\u5b9e\u73b0\u4e86\u5bf9\u7ec6\u80de\u8f6c\u5f55\u7ec4\u7684\u7edf\u4e00\u8868\u5f81\u3002<\/span><\/span><\/p>\n

\u5728\u591a\u4e2a\u4e0b\u6e38\u4efb\u52a1\u7684\u5b9e\u9a8c\u4e2d\uff0cscFoundation\u5c55\u73b0\u51fa\u4e86\u5353\u8d8a\u7684\u6027\u80fd\u3002\u5728\u57fa\u56e0\u8868\u8fbe\u589e\u5f3a\u4efb\u52a1\u4e0a\uff0c\u76f8\u6bd4\u4f20\u7edf\u65b9\u6cd5\u5c06\u5e73\u5747\u7edd\u5bf9\u8bef\u5dee\u964d\u4f4e\u4e86\u4e00\u534a\uff1b\u5728\u836f\u7269\u54cd\u5e94\u9884\u6d4b\u4e2d\uff0c\u65e0\u9700\u9488\u5bf9\u6027\u5fae\u8c03\u5c31\u8fbe\u5230\u4e86\u6700\u5148\u8fdb\u7684\u9884\u6d4b\u51c6\u786e\u7387\uff1b\u5728\u7ec6\u80de\u6270\u52a8\u9884\u6d4b\u4efb\u52a1\u4e2d\uff0c\u6a21\u578b\u80fd\u591f\u51c6\u786e\u9884\u6d4b\u57fa\u56e0\u6572\u9664\u540e\u7684\u8868\u8fbe\u53d8\u5316\u3002\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0cscFoundation\u901a\u8fc7\u9884\u8bad\u7ec3\u83b7\u5f97\u7684\u7ec6\u80de\u5d4c\u5165\u5411\u91cf\u548c\u57fa\u56e0\u4e0a\u4e0b\u6587\u5d4c\u5165\u5411\u91cf\uff0c\u5c55\u73b0\u51fa\u4e86\u4f18\u5f02\u7684\u751f\u7269\u5b66\u610f\u4e49\uff0c\u53ef\u4ee5\u7528\u4e8e\u7ec6\u80de\u7c7b\u578b\u6ce8\u91ca\u3001\u57fa\u56e0\u6a21\u5757\u53d1\u73b0\u7b49\u591a\u79cd\u5206\u6790\u4efb\u52a1\u3002<\/span><\/span><\/p>\n

\u901a\u8fc7\u5927\u89c4\u6a21\u9884\u8bad\u7ec3\uff0cscFoundation\u5b66\u4e60\u5230\u4e86\u666e\u904d\u7684\u751f\u7269\u5b66\u77e5\u8bc6\uff0c\u53ef\u4ee5\u8fc1\u79fb\u5230\u5404\u79cd\u4e0b\u6e38\u4efb\u52a1\u4e2d\u3002\u8fd9\u79cd\u65b9\u6cd5\u907f\u514d\u4e86\u4f20\u7edf\u65b9\u6cd5\u9700\u8981\u9488\u5bf9\u6bcf\u4e2a\u4efb\u52a1\u91cd\u65b0\u8bad\u7ec3\u6a21\u578b\u7684\u4f4e\u6548\u7387\uff0c\u4e3a\u751f\u547d\u79d1\u5b66\u7814\u7a76\u63d0\u4f9b\u4e86\u4e00\u4e2a\u7edf\u4e00\u7684\u8ba1\u7b97\u57fa\u7840\u8bbe\u65bd\u3002<\/span><\/span><\/p>\n

\u6b64\u5916\uff0c\u86cb\u767d\u8d28\u7ed3\u6784\u9884\u6d4b\u5728\u7406\u89e3\u86cb\u767d\u8d28\u7684\u7ed3\u6784-\u529f\u80fd\u5173\u7cfb\u3001\u63a8\u52a8\u751f\u7269\u5b66\u7814\u7a76\u4ee5\u53ca\u4fc3\u8fdb\u5236\u836f\u5f00\u53d1\u548c\u5b9e\u9a8c\u8bbe\u8ba1\u7b49\u65b9\u9762\u5177\u6709\u5173\u952e\u610f\u4e49\u3002\u5c3d\u7ba1\u65e5\u76ca\u6210\u719f\u7684\u6df1\u5ea6\u5b66\u4e60\u6280\u672f\u4ee5\u53ca\u4e30\u5bcc\u7684\u5b9e\u9a8c\u4e09\u7ef4\u86cb\u767d\u8d28\u7ed3\u6784\u6570\u636e\uff0c\u5728\u5f88\u5927\u7a0b\u5ea6\u4e0a\u52a0\u901f\u4e86\u7ed3\u6784\u9884\u6d4b\u7684\u53d1\u5c55\uff0c\u4f46\u86cb\u767d\u8d28\u7684\u7ed3\u6784\u52a8\u529b\u5b66\u7279\u6027\u5c1a\u672a\u5f97\u5230\u5145\u5206\u5173\u6ce8\u3002\u7136\u800c\uff0c\u86cb\u767d\u8d28\u5728\u529f\u80fd\u8fc7\u7a0b\u4e2d\u7684\u6784\u8c61\u53d8\u5316\u5bf9\u4e8e\u7cbe\u51c6\u836f\u7269\u8bbe\u8ba1\u81f3\u5173\u91cd\u8981\uff1a\u901a\u8fc7\u7eb3\u5165\u52a8\u6001\u53d8\u5316\u7684\u56e0\u7d20\uff0c\u80fd\u591f\u663e\u8457\u4f18\u5316\u836f\u7269\u4e0e\u86cb\u767d\u8d28\u7684\u7ed3\u5408\u6548\u7387\uff0c\u4ece\u800c\u63d0\u9ad8\u9776\u5411\u6cbb\u7597\u7684\u6709\u6548\u6027\u3002\u57fa\u4e8e\u6b64\uff0c\u6587\u732e[16]\u63d0\u51fa\u4e86\u4e00\u79cd\u521b\u65b0\u60274D\u6269\u6563\u6a21\u578b\uff0c\u8be5\u6a21\u578b\u7ed3\u5408\u4e86\u5206\u5b50\u52a8\u529b\u5b66<\/span>\uff08MD\uff09<\/span>\u6a21\u62df\u6570\u636e\uff0c\u65e8\u5728\u5b66\u4e60\u5e76\u6a21\u62df\u86cb\u767d\u8d28\u7684\u52a8\u6001\u7ed3\u6784\u3002\u8fd9\u4e00\u7814\u7a76\u5728\u884c\u4e1a\u5185\u7387\u5148\u5b9e\u73b0\u4e86\u5bf910\u7eb3\u79d2\u7ea7\u52a8\u6001\u8fd0\u52a8\u8fc7\u7a0b\u7684\u751f\u6210\uff0c\u4e3a\u86cb\u767d\u8d28\u7ed3\u6784\u52a8\u529b\u5b66\u5728\u836f\u7269\u8bbe\u8ba1\u7b49\u5b9e\u9645\u5e94\u7528\u573a\u666f\u4e2d\u7684\u8fdb\u4e00\u6b65\u63a2\u7d22\u63d0\u4f9b\u4e86\u91cd\u8981\u7684\u542f\u793a\u4e0e\u65b9\u6cd5\u652f\u6301\u3002<\/span><\/span><\/p>\n

\u533b\u5b66\u57fa\u7840\u6a21\u578b<\/span><\/span><\/h3>\n

\u63a8\u8350\u7406\u7531\uff1a<\/span><\/span><\/strong> GMAI (Generalist Medical AI)[9]\u4ee3\u8868\u4e86\u533b\u7597AI\u9886\u57df\u7684\u8303\u5f0f\u8f6c\u53d8\uff0c\u4ece\u4f20\u7edf\u7684\u4efb\u52a1\u7279\u5b9a\u6a21\u578b\u8f6c\u5411\u4e86\u66f4\u52a0\u901a\u7528\u548c\u7075\u6d3b\u7684\u57fa\u7840\u6a21\u578b\u65b9\u6cd5\u3002\u8fd9\u9879\u53d1\u8868\u5728Nature\u4e0a\u7684\u5f00\u521b\u6027\u5de5\u4f5c\u4e0d\u4ec5\u63d0\u51fa\u4e86\u533b\u7597AI\u7684\u65b0\u8303\u5f0f\uff0c\u66f4\u4e3a\u672a\u6765\u533b\u7597\u4fdd\u5065\u7684\u667a\u80fd\u5316\u63d0\u4f9b\u4e86\u5168\u65b0\u89c6\u89d2\u3002<\/span><\/span><\/p>\n

\u957f\u671f\u4ee5\u6765\uff0c<\/span><\/span>\u533b\u7597AI\u9762\u4e34\u7740\u4e09\u5927\u6838\u5fc3\u6311\u6218\uff1a\u6570\u636e\u7684\u7a00\u7f3a\u6027\u548c\u5f02\u8d28\u6027\u3001\u9884\u6d4b\u4efb\u52a1\u7684\u590d\u6742\u6027\u3001\u9886\u57df\u77e5\u8bc6\u7684\u6574\u5408\u96be\u5ea6\u3002\u4f20\u7edf\u7684\u533b\u7597AI\u7cfb\u7edf\u91c7\u7528\u201c\u4e00\u4e2a\u4efb\u52a1\u4e00\u4e2a\u6a21\u578b\u201d\u7684\u65b9\u5f0f\uff0c\u4e0d\u4ec5\u9700\u8981\u5927\u91cf\u6807\u6ce8\u6570\u636e\uff0c\u800c\u4e14\u96be\u4ee5\u5229\u7528\u6700\u65b0\u7684\u533b\u5b66\u77e5\u8bc6\u3002\u65e9\u671f\u7684\u4e00\u4e9b\u5c1d\u8bd5\uff0c\u5982CheXzero[17]\u5728\u653e\u5c04\u5f71\u50cf\u89e3\u8bfb\u4e0a\u7684\u63a2\u7d22\uff0c\u4ee5\u53caGPT-3[18]\u5728\u533b\u5b66\u6267\u7167\u8003\u8bd5\u4e0a\u7684\u8868\u73b0\uff0c\u90fd\u5c55\u793a\u4e86\u57fa\u7840\u6a21\u578b\u5728\u533b\u7597\u9886\u57df\u7684\u6f5c\u529b\uff0c\u4f46\u8ddd\u79bb\u4e34\u5e8a\u5e94\u7528\u4ecd\u6709\u4e0d\u5c0f\u5dee\u8ddd\u3002 <\/span><\/span><\/p>\n

GMAI\u7684\u521b\u65b0\u4e4b\u5904\u5728\u4e8e<\/span><\/span>\uff0c<\/span><\/span>\u5b83\u63d0\u51fa\u4e86\u4e00\u4e2a\u7edf\u4e00\u7684\u533b\u7597AI\u6846\u67b6\u3002\u8fd9\u4e2a\u6846\u67b6\u901a\u8fc7\u81ea\u76d1\u7763\u5b66\u4e60\u65b9\u5f0f\uff0c\u5728\u6d77\u91cf\u5f02\u6784\u533b\u7597\u6570\u636e\u4e0a\u8fdb\u884c\u9884\u8bad\u7ec3\uff0c\u540c\u65f6\u878d\u5165\u4e86\u7ed3\u6784\u5316\u7684\u533b\u5b66\u77e5\u8bc6\u3002\u5b83\u53ef\u4ee5\u8bad\u7ec3\u6765\u81ea\u56fe\u50cf\u3001EHR \u4fe1\u53f7\u3001\u7ec4\u5b66\u6570\u636e<\/span>\uff08\u57fa\u56e0\u7ec4\u5b66\u548c\u8f6c\u5f55\u7ec4\u5b66\uff09<\/span>\u7b49<\/span><\/span>\u7684<\/span><\/span>\u4e0d\u540c\u7c7b\u578b\u7684\u6570\u636e\uff0c\u80fd\u591f\u6444\u53d6\u533b\u5b66\u9886\u57df\u7684<\/span><\/span>\u591a\u6e90\u5f02\u6784<\/span><\/span>\u77e5\u8bc6<\/span><\/span>\uff08<\/span><\/span>\u51fa\u7248\u7269\u3001\u6559\u79d1\u4e66\u3001\u4e34\u5e8a\u8282\u70b9\u7684<\/span><\/span>\u77e5\u8bc6\u56fe\u8c31<\/span><\/span>\uff09<\/span><\/span>\uff0c\u8fd8\u5177\u6709\u96f6\u6837\u672c\u5b66\u4e60\u80fd\u529b\uff0c\u6839\u636e\u51e0\u4e2a\u4f8b\u5b50\u5c31\u53ef\u4ee5\u5b66\u4e60\u53bb\u505a\u6a21\u578b\u4ece\u6765\u6ca1\u6709\u89c1\u8fc7\u7684\u4efb\u52a1\u3002<\/span><\/span><\/p>\n

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

GMAI\uff08generalist medical AI\uff09\u6a21\u578b\u6d41\u7a0b\u56fe\u6982\u89c8 | \u6765\u6e90\uff1aMichael Moor\uff0c Oishi Banerjee\uff0c Zahra Shakeri Hossein Abad\uff0c et al. Foundation models for generalist medical artificial intelligence. nature\uff0c 2023\uff0c 616(7956): 259-265<\/span><\/span><\/em><\/span><\/p>\n

\u8fd9\u6837\u505a<\/span>\u7684\u597d\u5904\u662f<\/span><\/span>\uff0c<\/span><\/span><\/span>\u901a\u8fc7\u81ea\u7531\u7ec4\u5408\u548c\u4ea4\u9519\u4e0d\u540c\u7684\u6570\u636e\u6a21\u6001\uff0c\u80fd\u591f\u5b9e\u73b0\u52a8\u6001\u4efb\u52a1\u89c4\u8303<\/span>\uff08<\/span><\/span><\/span>D<\/span><\/span><\/span>ynamic <\/span><\/span><\/span>T<\/span><\/span><\/span>ask <\/span><\/span><\/span>S<\/span><\/span><\/span>pecification\uff09<\/span>\uff0c\u5373\u6839\u636e\u8fd9\u4e9b\u6307\u4ee4\u5373\u65f6\u89e3\u51b3\u4efb\u52a1\u3002\u56e0\u6b64\u57fa\u672c\u4e0a\u662f\u96f6\u6837\u672c\u6216\u5c11\u6837\u672c\u7684\u4e0a\u4e0b\u6587\u7c7b\u578b\u5b66\u4e60\u3002\u8fd9\u4e9b\u6a21\u578b\u5177\u6709\u5927\u91cf\u7684\u77e5\u8bc6\uff0c\u56e0\u4e3a\u5b83\u4eec\u5141\u8bb8\u6211\u4eec\u7528\u7ed3\u6784\u5316\u9886\u57df\u77e5\u8bc6\u8fdb\u884c\u63a8\u7406\u3002\u4f8b\u5982\uff0c\u5f53\u9047\u5230\u4e00\u4e2a\u590d\u6742\u7684\u75c5\u4f8b\u65f6\uff0c\u6a21\u578b\u4f1a\u9996\u5148\u5206\u6790\u5404\u9879\u68c0\u67e5\u7ed3\u679c\uff0c\u7136\u540e\u68c0\u7d22\u76f8\u5173\u7684\u533b\u5b66\u6587\u732e\u548c\u6307\u5357\uff0c\u6700\u540e\u901a\u8fc7\u903b\u8f91\u63a8\u7406\u7ed9\u51fa\u8bca\u65ad\u548c\u6cbb\u7597\u5efa\u8bae\uff0c\u5e76\u6e05\u6670\u5730\u89e3\u91ca\u63a8\u7406\u8fc7\u7a0b\u3002<\/span><\/span><\/span><\/p>\n

GMAI\u7684\u51fa\u73b0\u63a8\u52a8\u4e86\u4e00\u7cfb\u5217\u540e\u7eed\u521b\u65b0\u3002Med-Flamingo[19]\u57fa\u4e8eGMAI\u7684\u6846\u67b6\uff0c\u8fdb\u4e00\u6b65\u4f18\u5316\u4e86\u591a\u6a21\u6001\u533b\u7597\u8bca\u65ad\u80fd\u529b\uff1bRA-CM3[20]\u901a\u8fc7\u6539\u8fdb\u68c0\u7d22\u673a\u5236\uff0c\u663e\u8457\u63d0\u5347\u4e86\u6a21\u578b\u7684\u77e5\u8bc6\u83b7\u53d6\u80fd\u529b\uff1b\u800c\u57fa\u4e8eGMAI\u7684\u4e34\u5e8a\u8f85\u52a9\u7cfb\u7edf\u5df2\u7ecf\u5728\u591a\u5bb6\u533b\u9662\u5f00\u5c55\u8bd5\u70b9\u5e94\u7528\uff0c\u663e\u793a\u51fa\u826f\u597d\u7684\u5b9e\u9645\u6548\u679c\u3002<\/span><\/span><\/p>\n

\u8fd9\u4e9b\u9886\u57df\u4e5f\u6709\u5f88\u591a\u6311\u6218\uff0c\u5305\u62ec\u9a8c\u8bc1<\/span>\uff08Validation\uff09<\/span>\u3001\u786e\u8ba4<\/span>\uff08Verification\uff09<\/span>\u3001\u793e\u4f1a\u504f\u89c1<\/span>\uff08Social Biases\uff09<\/span>\u4ee5\u53ca\u975e\u5e38\u91cd\u8981\u7684\u60a3\u8005\u9690\u79c1\u95ee\u9898\u3002\u5373\u4fbf\u5982\u6b64\uff0c\u8fd9\u9879\u5de5\u4f5c\u7684\u610f\u4e49\u8fdc\u8d85\u6280\u672f\u521b\u65b0\uff0c\u5b83\u5f00\u542f\u4e86\u533b\u7597AI\u7684\u65b0\u7eaa\u5143\uff0c\u4e5f\u4e3a\u5176\u4ed6\u9886\u57df\u57fa\u7840\u6a21\u578b\u7684\u53d1\u5c55\u63d0\u4f9b\u4e86\u65b0\u7684\u601d\u8def\u3002<\/span><\/span><\/p>\n

\u6c14\u8c61\u9884\u6d4b\u57fa\u7840\u6a21\u578b<\/span><\/span><\/h3>\n

\u63a8\u8350\u7406\u7531\uff1a<\/span><\/span><\/strong> \u6c14\u8c61\u9884\u6d4b\u662f\u6700\u5feb\u80fd\u591f\u5728\u73b0\u5b9e\u751f\u6d3b\u5e94\u7528\u5e76\u5bf9\u751f\u4ea7\u751f\u6d3b\u4ea7\u751f\u5de8\u5927\u5f71\u54cd\u7684\u9886\u57df\u4e4b\u4e00\uff0c\u6c14\u8c61\u9884\u6d4b\u57fa\u7840\u6a21\u578b\u5728\u9884\u6d4b\u7cbe\u5ea6\u548c\u6548\u7387\u4e0a\u90fd\u5b9e\u73b0\u4e86\u7a81\u7834\u6027\u8fdb\u5c55\uff0c\u4e3a\u6781\u7aef\u5929\u6c14\u9884\u8b66\u548c\u6c14\u5019\u53d8\u5316\u7814\u7a76\u63d0\u4f9b\u4e86\u65b0\u5de5\u5177\u3002<\/span><\/span><\/p>\n

\u5982\u4f55\u5bf9\u672a\u6765\u5929\u6c14\u53d1\u5c55\u8fdb\u884c\u51c6\u786e\u9884\u6d4b\uff0c\u7279\u522b\u662f\u9884\u6d4b\u6781\u7aef\u5929\u6c14\uff0c\u8d8a\u6765\u8d8a\u6210\u4e3a\u610f\u4e49\u91cd\u5927\u7684\u79d1\u5b66\u95ee\u9898\u3002\u7136\u800c\uff0c\u4f20\u7edf\u7684\u6c14\u8c61\u5b66\u7814\u7a76\u53d7\u9650\u4e8e\u6570\u636e\u7684\u7f3a\u4e4f\u3001\u4eba\u7c7b\u77e5\u8bc6\u548c\u7ecf\u9a8c\u7684\u5c40\u9650\u4ee5\u53ca\u6c14\u8c61\u7cfb\u7edf\u5185\u5728\u7684\u4e0d\u786e\u5b9a\u6027\u56e0\u7d20\u7b49\u5f71\u54cd\uff0c\u65e0\u6cd5\u7ed9\u51fa\u8f83\u957f\u671f\u800c\u7cbe\u51c6\u7684\u9884\u6d4b\u3002<\/span><\/span><\/p>\n

\u8fd1\u5e74\u6765\uff0c\u82e5\u5e72\u6c14\u8c61\u9884\u6d4b\u5927\u6a21\u578b\u88ab\u9646\u7eed\u5f00\u53d1\u51fa\u6765\uff0c\u5305\u62ecGoogle\u7684GraphCast[21]\u3001\u534e\u4e3a\u7684\u76d8\u53e4\u5927\u6a21\u578b[10]\u3001\u82f1\u4f1f\u8fbe\u7684FourCastNet[22]\u3001\u4e0a\u6d77\u79d1\u5b66\u667a\u80fd\u7814\u7a76\u9662\u548c\u590d\u65e6\u5927\u5b66\u7684\u4f0f\u7fb2\u6c14\u8c61\u5927\u6a21\u578b\u7b49[11]<\/span><\/span>\uff0c<\/span><\/span>\u90fd\u53d6\u5f97\u4e86\u663e\u8457\u8fdb\u5c55\u3002<\/span><\/span><\/p>\n

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

 FuXi-S2S\u6a21\u578b\u7684\u7ed3\u6784\u793a\u610f\u56fe | Chen\uff0c L.\uff0c Zhong\uff0c X.\uff0c Zhang\uff0c F. et al. FuXi: a cascade machine learning forecasting system for 15-day global weather forecast. npj Clim Atmos Sci 6\uff0c 190 (2023).<\/span><\/span><\/em><\/span><\/p>\n

\u5168\u7403\u9996\u4e2a\u57fa\u4e8e\u4eba\u5de5\u667a\u80fd\u7684\u6b21\u5b63\u8282\u6c14\u5019\u9884\u6d4b\u5927\u6a21\u578b\u201c\u4f0f\u7fb2\u201d<\/span>\uff08FuXi-S2S)<\/span> \uff0c\u9996\u6b21\u8d85\u8d8a\u4f20\u7edf\u6570\u503c\u9884\u62a5\u6a21\u5f0f\u7684\u6807\u6746 \u2014\u2014 \u6b27\u6d32\u4e2d\u671f\u5929\u6c14\u9884\u62a5\u4e2d\u5fc3<\/span>\uff08ECMWF\uff09<\/span>\u7684 S2S \u6a21\u5f0f\u3002\u8be5\u6a21\u578b\u6db5\u76d6\u4e8613\u4e2a\u6c14\u538b\u5c42\u76845\u4e2a\u9ad8\u7a7a\u5927\u6c14\u53d8\u91cf\u548c11\u4e2a\u5730\u9762\u53d8\u91cf\uff0c\u5b9e\u73b0\u4e86\u4e24\u4e2a\u5173\u952e\u6027\u6280\u672f\u7a81\u7834\uff1a\u4e00\u662f\u5c06\u70ed\u5e26\u5927\u6c14\u5b63\u8282\u5185\u632f\u8361<\/span> (MJO) <\/span>\u8fd9\u4e00\u6b21\u5b63\u8282\u6700\u91cd\u8981\u7684\u53ef\u9884\u62a5\u6027\u6765\u6e90\u7eb3\u5165\u6a21\u578b\uff1b\u4e8c\u662f\u521b\u65b0\u6027\u5730\u8bbe\u8ba1\u4e86\u667a\u80fd\u6270\u52a8\u751f\u6210\u6a21\u5757\uff0c\u80fd\u591f\u5728\u9690\u7a7a\u95f4\u4e2d\u523b\u753b\u6c14\u5019\u7cfb\u7edf\u6f14\u53d8\u7684\u6982\u7387\u7279\u5f81\u3002\u8fd9\u4e9b\u521b\u65b0\u4f7f\u5f97\u6a21\u578b\u5728\u591a\u4e2a\u5173\u952e\u6307\u6807\u4e0a\u8d85\u8d8a\u4e86ECMWF S2S\u7cfb\u7edf\uff0c\u7279\u522b\u662f\u5c06MJO\u7684\u6709\u6548\u9884\u62a5\u671f\u4ece30\u5929\u5ef6\u957f\u523036\u5929\u3002<\/span><\/span><\/p>\n

\u5728\u5b9e\u9645\u5e94\u7528\u4e2d\uff0c\u201c\u4f0f\u7fb2\u201d\u4ec5\u97007\u79d2\u5373\u53ef\u5b8c\u621042\u5929\u7684\u5168\u7403\u65e5\u5e73\u5747\u9884\u62a5\uff0c\u5e76\u57282023\u5e74\u6210\u529f\u9884\u6d4b\u4e86\u5370\u5ea6\u5c3c\u897f\u4e9a\u7279\u5927\u66b4\u96e8\u4e8b\u4ef6\uff0c\u63d0\u524d20\u5929\u53d1\u51fa\u9884\u8b66\uff0c\u4e3a\u9632\u707e\u51cf\u707e\u8d62\u5f97\u4e86\u5b9d\u8d35\u65f6\u95f4\u3002\u6b64\u5916\uff0c\u590d\u65e6\u5927\u5b66\u8054\u5408\u4e0a\u6d77\u79d1\u5b66\u667a\u80fd\u7814\u7a76\u9662\u3001\u4e0a\u6d77\u521b\u667a\u5b66\u9662\u4e8e2024\u5e7411\u6708\u53d1\u5e03\u4e86\u4e2d\u56fd\u9996\u4e2a\u57fa\u4e8e\u81ea\u4e3b\u77e5\u8bc6\u4f53\u7cfb\u6784\u5efa\u7684\u6c14\u5019\u79d1\u5b66\u5927\u8bed\u8a00\u6a21\u578bPI@Climate[23]\uff0c\u8be5\u6a21\u578b\u5df2\u7ecf\u5b9e\u73b0\u4e86\u8de8\u8d8a46\u4e2a\u5b66\u79d1\u7684\u77e5\u8bc6\u6574\u5408\uff0c\u901a\u8fc7\u591a\u667a\u80fd\u4f53\u5efa\u6a21\u63d0\u5347\u4e86\u5bf9\u590d\u6742\u6c14\u5019\u7cfb\u7edf\u7684\u8ba4\u77e5\u80fd\u529b\u3002\u7269\u8d28\u79d1\u5b66\u57fa\u7840\u6a21\u578b<\/span><\/span><\/p>\n

\u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df<\/span><\/span><\/h4>\n

\u63a8\u8350\u7406\u7531\uff1a<\/span><\/span><\/strong>\u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df\u662f\u6750\u6599\u79d1\u5b66\u7814\u7a76\u7684\u57fa\u7840\u5de5\u5177\uff0c\u4f46\u957f\u671f\u53d7\u9650\u4e8e\u8ba1\u7b97\u6548\u7387\u4e0e\u7cbe\u5ea6\u7684\u6743\u8861\u95ee\u9898\u3002\u57fa\u4e8e\u6df1\u5ea6\u5b66\u4e60\u7684\u65b0\u578b\u529b\u573a\u6a21\u578b\u4e3a\u8fd9\u4e00\u96be\u9898\u63d0\u4f9b\u4e86\u7a81\u7834\u53e3\uff0c\u6709\u671b\u63a8\u52a8\u6750\u6599\u79d1\u5b66\u7814\u7a76\u8303\u5f0f\u7684\u9769\u65b0\u3002<\/span><\/span><\/p>\n

\u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df\u662f\u7814\u7a76\u6750\u6599\u5fae\u89c2\u884c\u4e3a\u7684\u91cd\u8981\u65b9\u6cd5\uff0c\u4f46\u5176\u51c6\u786e\u6027\u5f88\u5927\u7a0b\u5ea6\u4e0a\u4f9d\u8d56\u4e8e\u6240\u4f7f\u7528\u7684\u529b\u573a\u6a21\u578b\u3002\u4f20\u7edf\u7684\u89e3\u6790\u529b\u573a\u51fd\u6570\u5f62\u5f0f\u8f83\u4e3a\u7b80\u5355\uff0c\u96be\u4ee5\u51c6\u786e\u63cf\u8ff0\u590d\u6742\u7684\u539f\u5b50\u95f4\u76f8\u4e92\u4f5c\u7528\uff0c\u800c\u5bc6\u5ea6\u6cdb\u51fd\u7406\u8bba<\/span>\uff08DFT\uff09<\/span>\u867d\u7136\u51c6\u786e\u4f46\u8ba1\u7b97\u6210\u672c\u8fc7\u9ad8\uff0c\u8fd9\u4f7f\u5f97\u5927\u89c4\u6a21\u539f\u5b50\u4f53\u7cfb\u7684\u7cbe\u786e\u6a21\u62df\u9762\u4e34\u7740\u91cd\u5927\u6311\u6218\u3002<\/span><\/span><\/p>\n

\u9488\u5bf9\u8fd9\u4e9b\u95ee\u9898\uff0cGPTFF<\/span>\uff08Graph-based Pre-trained Transformer Force Field\uff09<\/span>[6]<\/span><\/span>\u4f5c\u4e3a<\/span><\/span>\u4e00\u79cd\u65b0\u578b\u7684\u901a\u7528\u4eba\u5de5\u667a\u80fd\u529b\u573a\u6a21\u578b\u88ab\u63d0\u51fa<\/span><\/span>\u3002<\/span><\/span>\u8be5\u6a21\u578b\u91c7\u7528\u56fe\u795e\u7ecf\u7f51\u7edc\u6765\u8868\u5f81\u6676\u4f53\u7ed3\u6784\uff0c\u5e76\u5f15\u5165<\/span><\/span>T<\/span><\/span>ransformer\u67b6\u6784\u6765\u63d0\u5347\u6a21\u578b\u7684\u5b66\u4e60\u80fd\u529b\u3002GPTFF\u6a21\u578b\u7684\u8bad\u7ec3\u6570\u636e\u6e90\u81ea\u81ea\u7814\u7684Atomly\u6750\u6599\u6570\u636e\u5e93<\/span>\uff08https:\/\/atomly.net\/\uff09<\/span>\u3002\u8be5\u6570\u636e\u5e93\u5305\u542b\u8fd135\u4e07\u4e2a\u65e0\u673a\u6750\u6599\u6570\u636e\uff0c\u6570\u636e\u89c4\u6a21\u548c\u8d28\u91cf\u4f4d\u4e8e\u4e16\u754c\u9876\u7ea7\u6c34\u51c6\uff0c\u77e5\u8bc6\u4ea7\u6743\u81ea\u4e3b\u53ef\u63a7\uff0c\u4e5f\u662f\u6211\u56fd\u552f\u4e00\u53ef\u5bf9\u6807\u6b27\u7f8e\u7ade\u54c1\u7684\u65e0\u673a\u6750\u6599\u6570\u636e\u5e93\u3002\u901a\u8fc7\u5229\u7528\u5305\u542b3780\u4e07\u4e2a\u5355\u70b9\u80fd\u91cf\u3001117\u4ebf\u4e2a\u539f\u5b50\u529b\u548c3.402\u4ebf\u4e2a\u5e94\u529b\u5f20\u91cf\u7684\u5927\u89c4\u6a21\u6570\u636e\u96c6\u8fdb\u884c\u8bad\u7ec3\uff0cGPTFF\u5728\u80fd\u91cf\u3001\u529b\u548c\u5e94\u529b\u9884\u6d4b\u65b9\u9762\u90fd\u8fbe\u5230\u4e86\u6781\u9ad8\u7684\u7cbe\u5ea6\uff0c\u5206\u522b\u5b9e\u73b0\u4e8632 meV\/atom\u300171 meV\/\u00c5\u548c0.365 GPa\u7684\u5e73\u5747\u7edd\u5bf9\u8bef\u5dee\uff0c\u6bd4\u4f20\u7edf\u7ecf\u9a8c\u529b\u573a\u63d0\u5347\u4e86\u4e00\u4e2a\u6570\u91cf\u7ea7\uff0c\u540c\u65f6\u8ba1\u7b97\u901f\u5ea6\u6bd4\u7b2c\u4e00\u6027\u539f\u7406\u5feb3-4\u4e2a\u6570\u91cf\u7ea7\u3002<\/span><\/span><\/p>\n

<\/section>\n

 GPTFF\u6a21\u578b\u7684\u6a21\u578b\u67b6\u6784| F. Xie\uff0c T. Lu\uff0c S. Meng\uff0c M. Liu\uff0c GPTFF: A high-accuracy out-of-the-box universal AI force field for arbitrary inorganic materials\uff0c Science Bulletin (2024)\uff0c doi: 10.1016\/j.scib.2024.08.039.<\/span><\/span><\/em><\/span><\/p>\n

\u4e0eGPTFF\u5e76\u9a7e\u9f50\u9a71\u7684\u8fd8\u6709DeepMD[24]\u3001PaiNN[25]\u548cUni-Mol+[26]\u7b49\u4ee3\u8868\u6027\u6a21\u578b\u3002DeepMD\u901a\u8fc7\u6df1\u5ea6\u52bf\u80fd\u6a21\u578b\u5b9e\u73b0\u4e86\u5bf9\u590d\u6742\u4f53\u7cfb\u7684\u51c6\u786e\u63cf\u8ff0\uff0c\u5df2\u6210\u529f\u5e94\u7528\u4e8e\u4e8c\u7ef4\u6750\u6599\u7684\u76f8\u53d8\u9884\u6d4b\uff1bPaiNN\u5219\u521b\u65b0\u6027\u5730\u5f15\u5165\u7b49\u53d8\u6027\u4fdd\u6301\u673a\u5236\uff0c\u5728\u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df\u4e2d\u5c55\u73b0\u51fa\u4f18\u5f02\u7684\u7a33\u5b9a\u6027\uff1bUni-Mol+\u521b\u65b0\u6027\u5730\u63d0\u51fa\u4e86\u57fa\u4e8e\u53cc\u8f68\u9053Transformer\u7684\u5206\u5b50\u8868\u793a\u65b9\u6cd5\uff0c\u901a\u8fc7\u8fed\u4ee3\u4f18\u5316\u521d\u59cb\u6784\u8c61\u6765\u903c\u8fd1DFT\u5e73\u8861\u6784\u8c61\uff0c\u5e76\u5728\u6b64\u57fa\u7840\u4e0a\u5b9e\u73b0\u91cf\u5b50\u5316\u5b66\u6027\u8d28\u9884\u6d4b\uff0c\u5728HOMO-LUMO\u80fd\u9699\u7b49\u5173\u952e\u5206\u5b50\u6027\u8d28\u9884\u6d4b\u4efb\u52a1\u4e0a\u5b9e\u73b0\u4e86\u663e\u8457\u8fdb\u5c55\u3002<\/span><\/span><\/p>\n

\u7269\u7406\u6a21\u62df<\/span><\/span><\/h4>\n

\u63a8\u8350\u7406\u7531\uff1a<\/span><\/span><\/strong>\u7269\u7406\u6a21\u62df\u662f\u79d1\u5b66\u7814\u7a76\u548c\u5de5\u7a0b\u5e94\u7528\u7684\u91cd\u8981\u5de5\u5177\uff0c\u4f46\u957f\u671f\u9762\u4e34\u7740\u4e0d\u540c\u7269\u7406\u7cfb\u7edf\u9700\u8981\u4e0d\u540c\u5efa\u6a21\u65b9\u6cd5\u7684\u6311\u6218\u3002Universal Physics Transformer<\/span> (UPT) <\/span>[7]\u901a\u8fc7\u521b\u65b0\u7684\u67b6\u6784\u8bbe\u8ba1\uff0c\u9996\u6b21\u5b9e\u73b0\u4e86\u5bf9\u4e0d\u540c\u7c7b\u578b\u7269\u7406\u7cfb\u7edf\u7684\u7edf\u4e00\u5904\u7406\uff0c\u4e3a\u53d1\u5c55\u7edf\u4e00\u7684\u7269\u7406\u6a21\u62df\u57fa\u7840\u6a21\u578b\u63d0\u4f9b\u4e86\u91cd\u8981\u8303\u5f0f\uff0c\u6709\u671b\u63a8\u52a8\u7269\u8d28\u79d1\u5b66\u7814\u7a76\u65b9\u6cd5\u7684\u9769\u65b0\u3002<\/span><\/span><\/p>\n

\u7269\u7406\u6a21\u62df\u5bf9\u63a8\u52a8\u6750\u6599\u79d1\u5b66\u7814\u7a76\u81f3\u5173\u91cd\u8981\uff0c\u4f46\u957f\u671f\u4ee5\u6765\u53d7\u56f0\u4e8e\u65b9\u6cd5\u7684\u788e\u7247\u5316\u95ee\u9898 \u2014\u2014\u4e0d\u540c\u7c7b\u578b\u7684\u7269\u7406\u7cfb\u7edf\u5f80\u5f80\u9700\u8981\u4e0d\u540c\u7684\u6a21\u578b\u548c\u7b97\u6cd5\uff0c\u8fd9\u4e0d\u4ec5\u589e\u52a0\u4e86\u7814\u7a76\u6210\u672c\uff0c\u4e5f\u9650\u5236\u4e86\u6a21\u62df\u6280\u672f\u7684\u5e7f\u6cdb\u5e94\u7528\u3002\u4f8b\u5982\uff0c\u5728\u6d41\u4f53\u52a8\u529b\u5b66\u6a21\u62df\u4e2d\uff0c\u5e38\u9700\u5728\u62c9\u683c\u6717\u65e5\u65b9\u6cd5\u4e0e\u6b27\u62c9\u65b9\u6cd5\u95f4\u6743\u8861\uff0c\u7f51\u683c\u6a21\u578b\u4e0e\u7c92\u5b50\u6a21\u578b\u4e5f\u5404\u6709\u4f18\u52a3\u3002\u8fd9\u79cd\u65b9\u6cd5\u7684\u788e\u7247\u5316\u4e0d\u4ec5\u589e\u52a0\u4e86\u6a21\u578b\u5f00\u53d1\u548c\u7ef4\u62a4\u7684\u6210\u672c\uff0c\u4e5f\u9650\u5236\u4e86\u7269\u7406\u6a21\u62df\u5728\u66f4\u5e7f\u6cdb\u573a\u666f\u4e0b\u7684\u5e94\u7528\u3002<\/span><\/span><\/p>\n

\u9488\u5bf9\u8fd9\u4e00\u95ee\u9898\uff0cUniversal Physics Transformer<\/span> (UPT) <\/span>\u63d0\u51fa\u4e86\u4e00\u4e2a\u7edf\u4e00\u7684\u795e\u7ecf\u7b97\u5b50\u6846\u67b6\u3002\u8be5\u6846\u67b6\u521b\u65b0\u6027\u5730\u91c7\u7528\u65e0\u7f51\u683c\/\u65e0\u7c92\u5b50\u7684\u6f5c\u5728\u8868\u793a\uff0c\u901a\u8fc7\u5206\u5c42Transformer\u67b6\u6784\u5b9e\u73b0\u4e86\u5bf9\u4e0d\u540c\u7c7b\u578b\u7269\u7406\u7cfb\u7edf\u7684\u7edf\u4e00\u5904\u7406\u3002UPT\u7684\u6838\u5fc3\u662f\u5176\u9ad8\u6548\u7684\u4e09\u6bb5\u5f0f\u8bbe\u8ba1\uff1a\u7f16\u7801\u5668\u7075\u6d3b\u5904\u7406\u4e0d\u540c\u7c7b\u578b\u7684\u8f93\u5165\u5e76\u538b\u7f29\u5230\u7edf\u4e00\u7684\u6f5c\u5728\u7a7a\u95f4\uff0c\u8fd1\u4f3c\u5668\u5728\u6f5c\u5728\u7a7a\u95f4\u4e2d\u8fdb\u884c\u52a8\u529b\u5b66\u4f20\u64ad\uff0c\u89e3\u7801\u5668\u5219\u652f\u6301\u5728\u4efb\u610f\u65f6\u7a7a\u70b9\u8fdb\u884c\u67e5\u8be2\u3002\u901a\u8fc7Inverse Encoding\/Decoding\u7b49\u521b\u65b0\u8bad\u7ec3\u7b56\u7565\uff0cUPT\u5b9e\u73b0\u4e86\u9ad8\u6548\u7684Latent Rollout\uff0c\u4f7f\u5176\u80fd\u591f\u6269\u5c55\u5230\u5927\u89c4\u6a21\u7269\u7406\u7cfb\u7edf\u7684\u6a21\u62df\u3002\u5728\u591a\u4e2a\u5177\u6709\u6311\u6218\u6027\u7684\u4efb\u52a1\u4e0a\uff0c\u5982\u7f51\u683c\u6d41\u4f53\u6a21\u62df\u3001Reynolds\u5e73\u5747Navier-Stokes\u6a21\u62df\u548c\u62c9\u683c\u6717\u65e5\u52a8\u529b\u5b66\u7b49\uff0cUPT\u90fd\u5c55\u73b0\u51fa\u4f18\u5f02\u7684\u6027\u80fd\uff0c\u540c\u65f6\u4fdd\u6301\u4e86\u826f\u597d\u7684\u53ef\u6269\u5c55\u6027\u548c\u6cdb\u5316\u80fd\u529b\u3002<\/span><\/span><\/p>\n

<\/section>\n

 UPT\u5b66\u4e60\u8303\u5f0f\u793a\u610f\u56fe\u67b6\u6784| B. Alkin\uff0c A. F\u00fcrst\uff0c S. Schmid\uff0c L. Gruber\uff0c M. Holzleitner and J. Brandstetter\uff0c “Universal Physics Transformers: A Framework For Efficiently Scaling Neural Operators\uff0c” in Proceedings of the 38th Conference on Neural Information Processing Systems (NeurIPS 2024)\uff0c 2024.<\/span><\/span><\/em><\/span><\/p>\n

UPT\u7684\u5de5\u4f5c\u4e3a\u6784\u5efa\u901a\u7528\u7684\u7269\u7406\u6a21\u62df\u57fa\u7840\u6a21\u578b\u63d0\u4f9b\u4e86\u91cd\u8981\u8303\u5f0f\u3002\u5b83\u4e0d\u4ec5\u5c55\u793a\u4e86\u5982\u4f55\u901a\u8fc7\u5de7\u5999\u7684\u67b6\u6784\u8bbe\u8ba1\u5b9e\u73b0\u4e0d\u540c\u7269\u7406\u7cfb\u7edf\u7684\u7edf\u4e00\u8868\u793a\u548c\u5904\u7406\uff0c\u8fd8\u4e3a\u63d0\u5347\u6a21\u578b\u53ef\u6269\u5c55\u6027\u63d0\u4f9b\u4e86\u65b0\u601d\u8def\u3002<\/span><\/span>\u8fd9\u5bf9\u4e8e\u53d1\u5c55\u66f4\u901a\u7528\u9ad8\u6548\u7684\u7269\u8d28\u79d1\u5b66\u6a21\u62df\u5de5\u5177\u5177\u6709\u91cd\u8981\u7684\u542f\u53d1\u610f\u4e49\u3002<\/span><\/span><\/p>\n


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

\n
\n
\n
\n
\n

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

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

\u6311\u6218\u4e0e\u5c55\u671b<\/span><\/strong><\/span><\/strong><\/strong><\/p>\n<\/section>\n

\n

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

<\/span><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n


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

\u76ee\u524d\uff0c\u79d1\u5b66\u5bb6\u4eec\u6b63\u5728\u5404\u4e2a\u9886\u57df\u63a2\u7d22\u57fa\u7840\u6a21\u578b\u7684\u5e94\u7528\u6f5c\u529b\uff0c\u540c\u65f6\u4e5f\u5728\u671d\u7740\u6784\u5efa\u7edf\u4e00\u79d1\u5b66\u57fa\u7840\u6a21\u578b\u7684\u5b8f\u4f1f\u76ee\u6807\u8fc8\u8fdb\u3002\u8fd9\u4e00\u8fdb\u7a0b\u9762\u4e34\u7740\u5f02\u6784\u6570\u636e\u878d\u5408\u3001\u7269\u7406\u89c4\u5f8b\u7ea6\u675f\u548c\u53ef\u89e3\u91ca\u6027\u7b49\u91cd\u8981\u6311\u6218\uff0c\u4f46\u8fd1\u671f\u7684\u7a81\u7834\u6027\u8fdb\u5c55\u5c55\u73b0\u4e86\u4ee4\u4eba\u632f\u594b\u7684\u524d\u666f\u3002<\/span><\/span><\/p>\n

\u5728\u6c14\u5019\u79d1\u5b66\u9886\u57df\uff0c\u7531\u590d\u65e6\u5927\u5b66\u7b49\u673a\u6784\u8054\u5408\u5f00\u53d1\u7684PI@Climate\u6a21\u578b[23]\u5df2\u7ecf\u5b9e\u73b0\u4e86\u8de8\u8d8a46\u4e2a\u5b66\u79d1\u7684\u77e5\u8bc6\u6574\u5408\uff0c\u901a\u8fc7\u591a\u667a\u80fd\u4f53\u5efa\u6a21\u63d0\u5347\u4e86\u5bf9\u590d\u6742\u6c14\u5019\u7cfb\u7edf\u7684\u8ba4\u77e5\u80fd\u529b\u3002\u5728\u751f\u547d\u79d1\u5b66\u9886\u57df\uff0c\u201c\u5973\u5a32\u201d\u7cfb\u5217\u6a21\u578b[23]\u5c55\u793a\u4e86\u4ece\u5fae\u89c2\u5230\u5b8f\u89c2\u7684\u7a81\u7834\uff1a\u5973\u5a32\u57fa\u56e0\u5bfc\u822a\u5927\u6a21\u578b\u901a\u8fc7\u56fe\u795e\u7ecf\u7f51\u7edc\u89e3\u5bc6\u4e86\u57fa\u56e0\u7ec4\u201c\u6697\u7269\u8d28\u201d\u7684\u4f5c\u7528\u673a\u5236\uff0c\u5973\u5a32\u751f\u547d\u6d41\u4f53\u5927\u6a21\u578b\u5b9e\u73b0\u4e86\u4eba\u4f53\u5185\u6d41\u4f53\u52a8\u529b\u5b66\u7684\u7cbe\u786e\u6a21\u62df\uff0c\u800c\u5973\u5a32\u751f\u7269\u7ed3\u6784\u5927\u6a21\u578b\u5219\u5728\u86cb\u767d\u8d28\u52a8\u6001\u7ed3\u6784\u9884\u6d4b\u65b9\u9762\u8d85\u8d8a\u4e86AlphaFold3\u7684\u6027\u80fd\u3002\u71e7\u4eba\u5206\u5b50\u57fa\u7840\u5927\u6a21\u578b[23]\u901a\u8fc7\u878d\u5408\u91cf\u5b50\u529b\u5b66\u6027\u8d28<\/span>\uff08\u5982\u80fd\u91cf\u3001\u529b\u3001\u7ed3\u6784\uff09<\/span>\u4e0e\u70ed\u529b\u5b66\u53c2\u6570<\/span>\uff08\u5305\u62ec\u7ed3\u6784\u5206\u5e03\u3001\u71b5\u3001\u7113\u53ca\u81ea\u7531\u80fd\u53d8\u5316\uff09<\/span>\uff0c\u5c06\u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df\u8fc7\u7a0b\u878d\u5165\u5927\u6a21\u578b\uff0c\u5b9e\u73b0\u4e86\u7269\u8d28\u5b8f\u89c2\u4e0e\u5fae\u89c2\u6027\u8d28\u7684\u5feb\u901f\u9884\u6d4b\uff0c\u5e76\u5728\u91cf\u5b50\u5316\u5b66\u6027\u8d28\u9884\u6d4b\u57fa\u51c6QM9\u6570\u636e\u96c6\u7684\u7edd\u5927\u591a\u6570\u4efb\u52a1\u4e2d\u8d85\u8d8a\u73b0\u6709AI\u6a21\u578b\u3002\u8fd9\u4e9b\u6210\u529f\u6848\u4f8b\u8bc1\u660e\uff0c\u5c06\u57fa\u7840\u6a21\u578b\u7406\u5ff5\u4e0e\u9886\u57df\u4e13\u4e1a\u77e5\u8bc6\u6df1\u5ea6\u878d\u5408\u662f\u53ef\u884c\u4e14\u5bcc\u6709\u6210\u6548\u7684\u3002<\/span><\/span><\/p>\n

\u968f\u7740\u6a21\u578b\u67b6\u6784\u7684\u521b\u65b0\u548c\u7b97\u529b\u7684\u63d0\u5347\uff0c\u6211\u4eec\u6709\u671b\u89c1\u8bc1\u4e00\u4e2a\u7edf\u4e00\u65b9\u6cd5\u8bba\u7684\u79d1\u5b66\u57fa\u7840\u6a21\u578b\u7684\u8bde\u751f\u3002\u8fd9\u4e2a\u6a21\u578b\u5c06\u4e0d\u518d\u5c40\u9650\u4e8e\u5355\u4e00\u5b66\u79d1\uff0c\u800c\u662f\u80fd\u591f\u7406\u89e3\u548c\u8fd0\u7528\u7269\u7406\u3001\u5316\u5b66\u3001\u751f\u7269\u7b49\u591a\u4e2a\u9886\u57df\u7684\u57fa\u672c\u89c4\u5f8b\uff0c\u5b9e\u73b0\u8de8\u5b66\u79d1\u7684\u77e5\u8bc6\u63a8\u7406\u548c\u521b\u65b0\u53d1\u73b0\u3002\u5b83\u5c06\u6210\u4e3a\u79d1\u5b66\u5bb6\u7684\u201c\u6570\u5b57\u667a\u56ca\u201d\uff0c\u4e0d\u4ec5\u80fd\u9884\u6d4b\u590d\u6742\u7cfb\u7edf\u7684\u884c\u4e3a\uff0c\u66f4\u80fd\u4e3b\u52a8\u63d0\u51fa\u7814\u7a76\u5047\u8bbe\u3001\u8bbe\u8ba1\u5b9e\u9a8c\u65b9\u6848\u3002\u5728\u6750\u6599\u7814\u53d1\u3001\u836f\u7269\u8bbe\u8ba1\u3001\u6c14\u5019\u9884\u6d4b\u7b49\u5173\u4e4e\u4eba\u7c7b\u798f\u7949\u7684\u91cd\u5927\u9886\u57df\uff0c\u8fd9\u79cd\u6a21\u578b\u5c06\u663e\u8457\u52a0\u901f\u79d1\u5b66\u53d1\u73b0\u7684\u8fdb\u7a0b<\/span><\/span><\/section>\n
\u53c2\u8003\u6587\u732e<\/span><\/span><\/section>\n

[1] Bommasani, Rishi, et al. “On the Opportunities and Risks of Foundation Models.” arXiv preprint arXiv:2108.07258 (2021).<\/span><\/span><\/p>\n

https:\/\/arxiv.org\/abs\/2108.07258<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u8fd9\u7bc7\u8bba\u6587\u4e3a\u6574\u4e2a\u62a5\u544a\u63d0\u4f9b\u4e86\u57fa\u7840\u6a21\u578b\u7684\u91cd\u8981\u80cc\u666f\u548c\u6982\u8ff0\uff0c\u4e3a\u540e\u7eed\u8ba8\u8bba\u5960\u5b9a\u4e86\u57fa\u7840\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[2] Abramson, J., <\/span><\/span>et al.<\/span><\/span><\/em> “Accurate Structure Prediction of Biomolecular Interactions with AlphaFold 3.” <\/span><\/span>Nature<\/span><\/span><\/em>630<\/span><\/span><\/strong>, 493\u2013500 (2024).<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41586-024-07487-w<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u8fd9\u662fAlphaFold\u7cfb\u5217\u6700\u65b0\u7684\u7a81\u7834\u6027\u6210\u679c\uff0c\u9996\u6b21\u5b9e\u73b0\u4e86\u7edf\u4e00\u7684\u751f\u7269\u5206\u5b50\u7ed3\u6784\u9884\u6d4b\u6846\u67b6\u3002\u8be5\u5de5\u4f5c\u5c06\u6df1\u5ea6\u5b66\u4e60\u5728\u7ed3\u6784\u751f\u7269\u5b66\u9886\u57df\u7684\u5e94\u7528\u63a8\u5411\u4e86\u65b0\u7684\u9ad8\u5ea6\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[3] Ramsundar, Bharath, et al. “Deep Learning for the Life Sciences.” O’Reilly Media (2019). https:\/\/www.amazon.com\/Deep-Learning-Life-Sciences-Microscopy\/dp\/1492039837<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aDeepChem\u88ab\u63d0\u53ca\u4e3a\u65e9\u671f\u5c06\u6df1\u5ea6\u5b66\u4e60\u5e94\u7528\u4e8e\u5206\u5b50\u6027\u8d28\u9884\u6d4b\u7684\u63a2\u7d22\uff0c\u63d0\u4f9b\u4e86\u91cd\u8981\u80cc\u666f\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[4] C. Wang, “A Review on 3D Convolutional Neural Network,” <\/span><\/span>2023 IEEE 3rd International Conference on Power, Electronics and Computer Applications (ICPECA)<\/span><\/span><\/em>, Shenyang, China, 2023, pp. 1204-1208,<\/span><\/span><\/p>\n

https:\/\/ieeexplore.ieee.org\/document\/10075760<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u4e09\u7ef4CNN\u88ab\u5f15\u7528\u4e3a\u6750\u6599\u7ed3\u6784\u5206\u6790\u7684\u65e9\u671f\u65b9\u6cd5\uff0c\u6709\u52a9\u4e8e\u8bf4\u660e\u8be5\u9886\u57df\u6280\u672f\u7684\u53d1\u5c55\u5386\u7a0b\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[5]Senior, Andrew W., et al. “Improved protein structure prediction using potentials from deep learning.” Nature 577.7792 (2020): 706-710.<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41586-019-1923-7<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aAlphaFold 1\u88ab\u6307\u51fa\u662f\u86cb\u767d\u8d28\u7ed3\u6784\u9884\u6d4b\u7684\u91cc\u7a0b\u7891\u5f0f\u7a81\u7834\uff0c\u662f\u4e86\u89e3\u8fd1\u671f\u8fdb\u5c55\u7684\u5173\u952e\u57fa\u7840\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[6] Xie, Feiyang, et al. “GPTFF: A high-accuracy out-of-the-box universal AI force field for arbitrary inorganic materials.” Science Bulletin (2024).<\/span><\/span><\/p>\n

https:\/\/www.sciencedirect.com\/science\/article\/pii\/S2095927324006327?via%3Dihub<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aGPTFF\u88ab\u5f3a\u8c03\u4e3a\u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df\u7684\u521b\u65b0\u529b\u573a\u6a21\u578b\uff0c\u4ee3\u8868\u4e86\u6750\u6599\u79d1\u5b66\u57fa\u7840\u6a21\u578b\u7684\u6700\u65b0\u91cd\u8981\u8fdb\u5c55\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[7] B. Alkin, A. F\u00fcrst, S. Schmid, L. Gruber, M. Holzleitner and J. Brandstetter, “Universal Physics Transformers: A Framework For Efficiently Scaling Neural Operators,” in Proceedings of the 38th Conference on Neural Information Processing Systems (NeurIPS 2024), 2024.<\/span><\/span><\/p>\n

https:\/\/openreview.net\/pdf?id=oUXiNX5KRm<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aUPT\u88ab\u4ecb\u7ecd\u4e3a\u4e00\u4e2a\u7edf\u4e00\u7684\u6846\u67b6\uff0c\u7528\u4e8e\u8de8\u4e0d\u540c\u7269\u7406\u7cfb\u7edf\u9ad8\u6548\u7f29\u653e\u795e\u7ecf\u7b97\u5b50\uff0c\u662f\u5c06\u57fa\u7840\u6a21\u578b\u5e94\u7528\u4e8e\u7269\u7406\u6a21\u62df\u7684\u91cd\u8981\u53d1\u5c55\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[8] Hao, Minsheng, et al. “Large-scale foundation model on single-cell transcriptomics.” Nature Methods 21.12 (2024): 1481-1491.<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41592-024-02305-7<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1ascFoundation\u88ab\u805a\u7126\u4e3a\u5c06\u57fa\u7840\u6a21\u578b\u5e94\u7528\u4e8e\u5355\u7ec6\u80de\u8f6c\u5f55\u7ec4\u6570\u636e\u7684\u7a81\u7834\uff0c\u514b\u670d\u4e86\u5173\u952e\u6311\u6218\uff0c\u4e3a\u8be5\u9886\u57df\u5f00\u8f9f\u4e86\u65b0\u7684\u53ef\u80fd\u6027\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[9] Moor, Michael, et al. “Foundation models for generalist medical artificial intelligence.” Nature 616.7956 (2023): 259-265.<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41586-023-05881-4<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aGMAI\u88ab\u5448\u73b0\u4e3a\u5f00\u521b\u6027\u7684\u901a\u7528\u533b\u7597AI\u6a21\u578b\uff0c\u4ee3\u8868\u4e86\u533b\u7597AI\u7684\u8303\u5f0f\u8f6c\u53d8\uff0c\u4e3a\u667a\u80fd\u533b\u7597\u7684\u672a\u6765\u63d0\u4f9b\u4e86\u65b0\u89c6\u89d2\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[10] Bi, K., Xie, L., Zhang, H. <\/span><\/span>et al.<\/span><\/span><\/em> Accurate medium-range global weather forecasting with 3D neural networks. <\/span><\/span>Nature<\/span><\/span><\/em>619<\/span><\/span><\/strong>, 533\u2013538 (2023).<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41586-023-06185-3<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u9762\u541120\u65e5\u5168\u7403\u5929\u6c14\u9884\u62a5\u7684\u9ad8\u5206\u8fa8\u7387\u901a\u7528\u6a21\u578b\u76d8\u53e4\u5929\u6c14\u9884\u62a5\u6a21\u578b\u662f\u57fa\u7840\u6a21\u578b\u5728\u5929\u6c14\u9884\u62a5\u9886\u57df\u7684\u4ee3\u8868\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[11] Chen, L., Zhong, X., Zhang, F. <\/span><\/span>et al.<\/span><\/span><\/em> FuXi: a cascade machine learning forecasting system for 15-day global weather forecast. <\/span><\/span>npj Clim Atmos Sci<\/span><\/span><\/em>6<\/span><\/span><\/strong>, 190 (2023).<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41612-023-00512-1<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aFuXi\u662f\u4e00\u4e2a\u7528\u4e8e15\u5929\u5168\u7403\u5929\u6c14\u9884\u62a5\u7684\u7ea7\u8054\u673a\u5668\u5b66\u4e60\u9884\u62a5\u7cfb\u7edf\uff0c\u5c55\u73b0\u4e86\u5929\u6c14\u9884\u62a5\u9886\u57df\u7684\u6700\u65b0\u8fdb\u5c55\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[12] Kang Chen, Tao Han, Junchao Gong, et al. <\/span><\/span>FengWu: Pushing the Skillful Global Medium-range Weather Forecast beyond 10 Days Lead<\/span><\/span><\/strong>. arXiv:2304.02948<\/span><\/span><\/em>, 2023<\/span><\/span><\/p>\n

https:\/\/arxiv.org\/abs\/2304.02948<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aFengWu\u662f\u4e00\u79cd\u5148\u8fdb\u7684AI\u9a71\u52a8\u5168\u7403\u4e2d\u671f\u5929\u6c14\u9884\u62a5\u7cfb\u7edf\uff0c\u901a\u8fc7\u591a\u6a21\u6001\u4e0e\u591a\u4efb\u52a1\u5b66\u4e60\u65b9\u6cd5\u663e\u8457\u63d0\u5347\u4e86\u5929\u6c14\u9884\u62a5\u7684\u7cbe\u5ea6\u548c\u65f6\u6548\u6027<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[13]Chen, Kang, et al. “An Overview of Domain-Specific Foundation Models: Key Technologies, Applications, and Challenges.” <\/span><\/span>arXiv<\/span><\/span><\/em>, 2024, arXiv:2409.04267,<\/span><\/span><\/p>\n

https:\/\/arxiv.org\/abs\/2409.04267<\/span><\/span><\/span>.<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u8fd9\u7bc7\u6587\u7ae0\u5168\u9762\u6982\u8ff0\u4e86\u5b9a\u5236\u5316\u9886\u57df\u4e13\u7528\u57fa\u7840\u6a21\u578b\u7684\u65b9\u6cd5\u8bba\uff0c\u4ecb\u7ecd\u4e86\u57fa\u7840\u6982\u5ff5\u3001\u901a\u7528\u67b6\u6784\uff0c\u5e76\u603b\u7ed3\u4e86\u6784\u5efa\u9886\u57df\u4e13\u7528\u6a21\u578b\u7684\u5173\u952e\u65b9\u6cd5\u4ee5\u53ca\u9762\u4e34\u7684\u6311\u6218\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[14] Theodoris, C.V., Xiao, L., Chopra, A. <\/span><\/span>et al.<\/span><\/span><\/em> Transfer learning enables predictions in network biology. <\/span><\/span>Nature<\/span><\/span><\/em>618<\/span><\/span><\/strong>, 616\u2013624 (2023).<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41586-023-06139-9<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u57fa\u56e0\u8868\u8fbe\u57fa\u7840\u6a21\u578b\u5efa\u6a21\u7684\u5148\u9a71\u5de5\u4f5cGeneformer<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[15] Cui, H., Wang, C., Maan, H. <\/span><\/span>et al.<\/span><\/span><\/em> scGPT: toward building a foundation model for single-cell multi-omics using generative AI. <\/span><\/span>Nat Methods<\/span><\/span><\/em>21<\/span><\/span><\/strong>, 1470\u20131480 (2024).<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41592-024-02201-0<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u57fa\u56e0\u8868\u8fbe\u57fa\u7840\u6a21\u578b\u5efa\u6a21\u7684\u91cd\u8981\u5de5\u4f5cscGPT<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[16]<\/span><\/span>K. Cheng, C. Liu, Q. Su, J. Wang, L. Zhang, Y. Tang, Y. Yao, S. Zhu, Y. Qi. 4D Diffusion for Dynamic Protein Structure Prediction with Reference Guided Motion Alignment. AAAI 2025.<\/span><\/span><\/span><\/p>\n

\u63a8\u8350\u7406\u7531\uff1a\u884c\u4e1a\u9996\u6b21\u5b9e\u73b0\u751f\u621010\u7eb3\u79d2\u7ea7\u7684\u52a8\u6001\u8fd0\u52a8\u8fc7\u7a0b\u3002<\/span><\/span><\/span><\/p>\n

[17] Tiu, E., Talius, E., Patel, P. <\/span><\/span>et al.<\/span><\/span><\/em> Expert-level detection of pathologies from unannotated chest X-ray images via self-supervised learning. <\/span><\/span>Nat. Biomed. Eng<\/span><\/span><\/em>6<\/span><\/span><\/strong>, 1399\u20131406 (2022).<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41551-022-00936-9<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aCheXzero\u88ab\u6307\u51fa\u662f\u5728\u653e\u5c04\u5b66\u89e3\u8bfb\u4e2d\u5e94\u7528\u533b\u7597\u4eba\u5de5\u667a\u80fd\u7684\u65e9\u671f\u63a2\u7d22\uff0c\u6709\u52a9\u4e8e\u8bf4\u660e\u5c06\u57fa\u7840\u6a21\u578b\u5e94\u7528\u4e8e\u533b\u5b66\u7684\u6f5c\u529b\u548c\u6311\u6218\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[18] Brown, Tom, et al. “Language models are few-shot learners.” Advances in neural information processing systems 33 (2020): 1877-1901. https:\/\/proceedings.neurips.cc\/paper\/2020\/hash\/1457c0d6bfcb4967418bfb8ac142f64a-Abstract.html<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aGPT-3\u5728\u533b\u5e08\u8d44\u683c\u8003\u8bd5\u4e2d\u7684\u8868\u73b0\uff0c\u88ab\u5f15\u7528\u4e3a\u57fa\u7840\u6a21\u578b\u5728\u533b\u5b66\u9886\u57df\u6f5c\u529b\u7684\u4e00\u4e2a\u6307\u793a\uff0c\u4f7f\u5176\u6210\u4e3a\u8be5\u90e8\u5206\u7684\u76f8\u5173\u53c2\u8003\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[19] Moor, Michael, et al. “Med-Flamingo: A Multimodal Medical Few-shot Learner.” <\/span><\/span>arXiv<\/span><\/span><\/em>, July 2023, arXiv:2307.15189,<\/span><\/span><\/p>\n

https:\/\/arxiv.org\/abs\/2307.15189<\/span><\/span><\/span>.<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aMed-Flamingo\u6a21\u578b\u5728\u591a\u6a21\u6001\u533b\u7597\u8bca\u65ad\u65b9\u9762\u8fdb\u884c\u4e86\u4f18\u5316\uff0c\u5ef6\u7eed\u4e86GMAI\u5de5\u4f5c\u7684\u521b\u65b0\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[20] Chen, Xiaoli, et al. “Retrieval-Augmented Multimodal Language Modeling” arXiv preprint arXiv:2212.06759 (2022).<\/span><\/span><\/p>\n

https:\/\/arxiv.org\/abs\/2211.12561<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aRA-CM3\u662f\u53e6\u4e00\u4e2a\u57fa\u4e8eGMAI\u7684\u6a21\u578b\uff0c\u5177\u6709\u6539\u8fdb\u7684\u77e5\u8bc6\u68c0\u7d22\u80fd\u529b\uff0c\u5c55\u793a\u4e86\u8fd1\u671f\u533b\u7597\u4eba\u5de5\u667a\u80fd\u7684\u65b0\u8fdb\u5c55\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[21] Remi Lam et al., Learning skillful medium-range global weather forecasting.Science<\/span><\/span>382<\/span><\/span><\/strong>,1416-1421(2023)<\/span><\/span><\/p>\n

https:\/\/www.science.org\/doi\/10.1126<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aGraphCast\u662f\u8c37\u6b4c\u7684\u4e00\u4e2a\u5929\u6c14\u9884\u6d4b\u6a21\u578b\uff0c\u88ab\u4f5c\u4e3a\u4eba\u5de5\u667a\u80fd\u5929\u6c14\u9884\u62a5\u7cfb\u7edf\u7684\u4ee3\u8868\u5f15\u7528\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[22] Jaderberg, Max, et al. “FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators.” arXiv preprint arXiv:2202.11214 (2022).<\/span><\/span><\/p>\n

https:\/\/arxiv.org\/abs\/2202.11214<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1aFourCastNet\u662f\u82f1\u4f1f\u8fbe\u7684\u4e00\u4e2a\u5929\u6c14\u9884\u6d4b\u6a21\u578b\uff0c\u57fa\u4e8e\u9002\u5e94\u6027\u5085\u91cc\u53f6\u795e\u7ecf\u7b97\u5b50\uff0c\u63d0\u4f9b\u4e86\u5929\u6c14\u9884\u62a5\u6a21\u578b\u6784\u5efa\u7684\u6709\u76ca\u53c2\u8003\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[23]\u201c\u4e0a\u667a\u9662\u53d1\u5e03\u7cfb\u5217\u5782\u76f4\u9886\u57df\u79d1\u5b66\u5927\u6a21\u578b\u201d <\/span><\/span>\u79d1\u5b66\u7f51\uff0c<\/span><\/span><\/em>2024\u5e7411\u6708\uff0c<\/span><\/span>https:\/\/news.sciencenet.cn\/htmlnews\/2024\/11\/533897.shtm<\/span><\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u4e0a\u6d77\u79d1\u5b66\u667a\u80fd\u7814\u7a76\u9662\u53d1\u5e03\u7684\u4e00\u7cfb\u5217\u57fa\u7840\u6a21\u578b\u4ecb\u7ecd\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[24] Zhang, L., et al. “Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics.” Physical Review Letters (2018).<\/span><\/span><\/p>\n

https:\/\/journals.aps.org\/prl\/abstract\/10.1103\/PhysRevLett.120.143001<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u8fd9\u662f\u6df1\u5ea6\u5b66\u4e60\u5728\u5206\u5b50\u52a8\u529b\u5b66\u9886\u57df\u7684\u5960\u57fa\u6027\u5de5\u4f5c\uff0c\u9996\u6b21\u5b9e\u73b0\u4e86\u91cf\u5b50\u529b\u5b66\u7cbe\u5ea6\u7684\u5927\u89c4\u6a21\u6a21\u62df\u3002\u8be5\u65b9\u6cd5\u4e3a\u6750\u6599\u79d1\u5b66\u7814\u7a76\u63d0\u4f9b\u4e86\u5f3a\u5927\u7684\u8ba1\u7b97\u5de5\u5177\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[25] Sch\u00fctt, K., et al. “Equivariant Message Passing for the Prediction of Tensorial Properties and Molecular Spectra.” arXiv preprint arXiv:2102.03150 (2021).<\/span><\/span><\/p>\n

https:\/\/arxiv.org\/abs\/2102.03150<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u8fd9\u9879\u5de5\u4f5c\u5728\u5206\u5b50\u6027\u8d28\u9884\u6d4b\u4e2d\u5f15\u5165\u4e86\u7b49\u53d8\u6027\u673a\u5236\uff0c\u663e\u8457\u63d0\u5347\u4e86\u9884\u6d4b\u7cbe\u5ea6\u3002\u8be5\u65b9\u6cd5\u4e3a\u5206\u5b50\u52a8\u529b\u5b66\u6a21\u62df\u63d0\u4f9b\u4e86\u65b0\u7684\u7406\u8bba\u6846\u67b6\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

[26] Lu, S., Gao, Z., He, D. <\/span><\/span>et al.<\/span><\/span><\/em> Data-driven quantum chemical property prediction leveraging 3D conformations with Uni-Mol+. <\/span><\/span>Nat Commun<\/span><\/span><\/em>15<\/span><\/span><\/strong>, 7104 (2024).<\/span><\/span><\/p>\n

https:\/\/www.nature.com\/articles\/s41467-024-51321<\/span><\/span><\/p>\n\n\n\n
\n

\u63a8\u8350\u7406\u7531<\/span><\/span><\/strong><\/span>\uff1a\u8fd9\u9879\u5de5\u4f5c\u63d0\u51fa\u4e86\u521b\u65b0\u7684\u53cc\u8f68\u9053Transformer\u5206\u5b50\u8868\u793a\u65b9\u6cd5\uff0c\u5b9e\u73b0\u4e86\u9ad8\u7cbe\u5ea6\u7684\u5206\u5b50\u6027\u8d28\u9884\u6d4b\u3002\u8be5\u6a21\u578b\u5728\u591a\u4e2a\u5316\u5b66\u6027\u8d28\u9884\u6d4b\u4efb\u52a1\u4e0a\u90fd\u53d6\u5f97\u4e86\u7a81\u7834\u6027\u8fdb\u5c55\u3002<\/span><\/span><\/span><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n


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

\n

\u51fa\u54c1<\/span><\/span><\/strong>\uff1a\u6f06\u8fdc\u3001\u5434\u529b\u6ce2\u3001\u5f20\u6c5f<\/span><\/span><\/span><\/p>\n

\u8fd0\u8425<\/span><\/span><\/strong>\uff1a\u5b5f\u664b\u5b87\u3001\u738b\u5a77<\/span><\/span><\/p>\n

\u64b0\u7a3f<\/span><\/span><\/strong>\uff1a\u5f20\u6c5f\u3001\u6768\u71d5\u9752\u3001\u738b\u5a77\u3001\u738b\u671d\u4f1a\u3001\u5341\u4e09\u7ef4\u3001\u5468\u8389\u3001\u6881\u91d1\u3001\u8881\u51b0\u3001\u6c5f\u5343\u6708\u3001\u5218\u5fd7\u6bc5<\/span><\/span><\/p>\n

\u9e23\u8c22<\/span><\/span><\/strong>\uff08\u6309\u59d3\u6c0f\u62fc\u97f3\u987a\u5e8f\uff0c\u6392\u540d\u4e0d\u5206\u5148\u540e\uff09\uff1a<\/span><\/span><\/p>\n

\u66f9\u98ce\u96f7 \u3001\u9648\u5c0f\u6768 \u3001\u7a0b\u8fdc\u3001\u675c\u6c85\u5c82 \u3001\u6bb5\u90c1\u3001\u65b9\u69af\u6977 \u3001\u4ed8\u5f66\u4f1f\u3001 \u9ad8\u60a6\u3001\u9ec4\u67ef\u946b\u3001\u674e\u660a\u3001\u5218\u5723\u8d85\u3001\u8c2d\u4f1f\u654f\u3001\u5434\u6cf0\u9716\u3001\u5434\u8273\u73b2\u3001\u5411\u7ea2\u519b\u3001\u5f20\u9aa5\u3001\u5f20\u8273\u3001\u6731\u601d\u8bed<\/span><\/span><\/p>\n<\/blockquote>\n


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


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

<\/section>\n


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

\n
\n
\n
\n

AI+Science \u8bfb\u4e66\u4f1a<\/span><\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\n


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

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

\"\u56fe\u7247\"<\/span><\/a><\/p>\n

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

<\/span><\/section>\n
\u8be6\u60c5\u8bf7\u89c1\uff1a<\/span><\/span><\/span><\/section>\n
\u4eba\u5de5\u667a\u80fd\u548c\u79d1\u5b66\u53d1\u73b0\u76f8\u4e92\u8d4b\u80fd\u7684\u65b0\u8303\u5f0f\uff1aAI+Science \u8bfb\u4e66\u4f1a\u542f\u52a8<\/a><\/span><\/span><\/span><\/section>\n

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

<\/span><\/section>\n
\u63a8\u8350\u9605\u8bfb<\/span><\/span><\/strong><\/strong><\/strong><\/section>\n
<\/strong><\/section>\n
1. \u300aAI X Science\u5341\u5927\u524d\u6cbf\u89c2\u5bdf\u300b\u53d1\u5e03\uff1a35\u4e2a\u7814\u7a76\u524d\u6cbf\uff0c\u5f00\u542f\u79d1\u5b66\u53d1\u73b0\u7684\u9ec4\u91d1\u65f6\u4ee3<\/span><\/a><\/span><\/strong>
<\/span><\/strong><\/span><\/section>\n
2. \u300aAI\u00d7SCIENCE\u5341\u5927\u524d\u6cbf\u89c2\u5bdf\u300b1\uff1a\u57fa\u4e8eLLM\u7684\u79d1\u5b66\u7814\u7a76<\/span><\/strong>
<\/span><\/strong><\/span><\/section>\n
3. AI\u79d1\u5b66\u5bb6\uff1a\u81ea\u52a8\u53d1\u73b0\u7269\u7406\u7cfb\u7edf\u7684\u9690\u85cf\u72b6\u6001\u53d8\u91cf<\/span><\/a><\/span><\/strong><\/section>\n
<\/strong><\/section>\n
4. <\/span><\/strong><\/span>\u6e38\u620f\u5316\u79d1\u7814\u2014\u2014\u8ba9\u6211\u4eec\u7a81\u7834\u5185\u5377\u3001\u5171\u7814\u6d8c\u73b0<\/a><\/span><\/strong><\/strong><\/span><\/section>\n
<\/strong>5. <\/span><\/strong>\u63a2\u7d22\u8005\u8ba1\u5212 | \u96c6\u667a\u4ff1\u4e50\u90e82025\u5185\u5bb9\u56e2\u961f\u62db\u52df\uff08\u5168\u804c&\u517c\u804c\uff09<\/a><\/span><\/strong><\/span><\/section>\n

6. \u52a0\u5165\u96c6\u667a\uff0c\u73a9\u8f6c\u590d\u6742\uff0c\u5171\u521b\u6591\u56fe\uff01\u96c6\u667a\u4ff1\u4e50\u90e8\u7ebf\u4e0b\u5fd7\u613f\u8005\u62db\u52df<\/span><\/a><\/span><\/strong>
<\/span><\/strong><\/span><\/p>\n


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


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

\u70b9\u51fb\u201c\u9605\u8bfb\u539f\u6587\u201d\uff0c\u62a5\u540d\u8bfb\u4e66\u4f1a<\/span><\/span><\/section>\n<\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

\u5bfc\u8bed \u4e0a\u6d77\u79d1\u5b66\u667a\u80fd\u7814\u7a76\u9662\u3001\u96c6\u667a\u79d1\u5b66\u7814\u7a76\u4e2d\u5fc3\u548c\u963f\u91cc\u4e91\u8054\u5408\u53d1\u5e03\u4e86\u300aAI X Science\u5341\u5927\u524d\u6cbf\u89c2\u5bdf\u300b\uff0c\u68b3\u7406\u51fa35\u4e2a\u7814\u7a76\u524d\u6cbf\uff0c\u6765\u63a8\u52a8\u79d1\u5b66\u53d1\u5c55\u7684\u9ec4\u91d1\u65f6\u4ee3\u5230\u6765\u3002\u672c\u7bc7\u4e3a\u524d\u6cbf\u89c2\u5bdf2\uff0c\u626b\u63cf\u4e0b\u65b9\u4e8c\u7ef4\u7801\uff0c\u53ef\u83b7\u5f97\u5b8c\u6574\u7248\u4e0b\u8f7d\u5730\u5740\uff0c\u5e76\u80fd\u5feb\u901f\u94fe\u63a5\u8bba\u6587\u539f\u6587\u3002 \u626b\u63cf\u4e8c\u7ef4\u7801 \u4e0b\u8f7d\u5b8c\u6574\u62a5\u544a \u5782\u76f4\u9886\u57df\u79d1\u5b66\u5927\u6a21\u578b \u80cc\u666f\u4ecb\u7ecd \u4eba\u5de5\u667a\u80fd\u7684…<\/p>\n","protected":false},"author":0,"featured_media":59040,"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\/59053"}],"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=59053"}],"version-history":[{"count":0,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/59053\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/media\/59040"}],"wp:attachment":[{"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=59053"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=59053"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=59053"},{"taxonomy":"special","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fspecial&post=59053"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}