{"id":21925,"date":"2020-11-15T08:42:49","date_gmt":"2020-11-15T00:42:49","guid":{"rendered":"https:\/\/swarma.org\/?p=21925"},"modified":"2020-11-15T08:42:49","modified_gmt":"2020-11-15T00:42:49","slug":"pm2-5-gnn-%e5%9b%be%e7%bd%91%e7%bb%9c%e9%9b%be%e9%9c%be%e9%a2%84%e6%8a%a5%e6%a8%a1%e5%9e%8b%ef%bc%9a%e5%a4%8d%e6%9d%82%e7%b3%bb%e7%bb%9f%e8%87%aa%e5%8a%a8%e5%bb%ba%e6%a8%a1%e5%9c%a8%e5%a4%a7%e6%b0%94","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=21925","title":{"rendered":"PM2.5-GNN \u56fe\u7f51\u7edc\u96fe\u973e\u9884\u62a5\u6a21\u578b\uff1a\u590d\u6742\u7cfb\u7edf\u81ea\u52a8\u5efa\u6a21\u5728\u5927\u6c14\u6c61\u67d3\u4e2d\u7684\u5e94\u7528"},"content":{"rendered":"
\n
<\/section>\n

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

<\/section>\n
\u5bfc\u8bed<\/span><\/strong><\/span><\/section>\n
\n
\n
\n
\u672c\u6587\u603b\u7ed3\u4e86\u5317\u4eac\u5e08\u8303\u5927\u5b66\u7cfb\u7edf\u79d1\u5b66\u5b66\u9662\u5728\u8bfb\u535a\u58eb\u751f\u738b\u7855\u5728\u96c6\u667a-\u51ef\u98ce\u7814\u8bfb\u8425\u4e0a\u7684\u5206\u4eab\uff0c\u4ecb\u7ecd\u5176\u53d1\u8868\u4e8e SIGSPATIAL 2020 \u7684\u5de5\u4f5c\u300aPM2.5<\/sub>-GNN: A Domain Knowledge Enhanced Graph Neural Network For PM<\/span>2.5<\/sub> Forecasting\u300b\uff08\u878d\u5408\u9886\u57df\u77e5\u8bc6\u7684\u56fe\u795e\u7ecf\u7f51\u7edc\u96fe\u973e\u9884\u62a5\u6a21\u578b\uff09\u3002\u8fd9\u9879\u5de5\u4f5c\u5c06\u56fe\u795e\u7ecf\u7f51\u7edc\uff08Graph Neural Network\uff0cGNN\uff09\u4e0e\u5faa\u73af\u795e\u7ecf\u7f51\u7edc\uff08GRU\uff0cGated Recurrent Units\uff09\u7ed3\u5408\uff0c\u5e76\u878d\u5165\u5927\u91cf\u9886\u57df\u5148\u9a8c\u77e5\u8bc6\uff0c\u5728\u57fa\u4e8e\u56fe\uff08Graph\uff09\u5efa\u6a21\u7684 PM2.5<\/sub> \u9884\u62a5\u4e2d\u8fbe\u5230\u4e86\u5f53\u524d\u6700\u4f18\uff08state-of-the-art, SOTA\uff09\u7684\u6548\u679c\u3002<\/span><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n


<\/p>\n

\n
\n
\n

\u8bba\u6587\u9898\u76ee\uff1a<\/span><\/p>\n

PM<\/span>2.5<\/sub>-GNN: A Domain Knowledge Enhanced Graph Neural Network For P<\/span>M<\/span>2.5<\/sub> Forecasting<\/span><\/span><\/p>\n

\u8bba\u6587\u5730\u5740\uff1a<\/span><\/p>\n

https:\/\/arxiv.org\/abs\/2002.12898v1<\/span><\/p>\n<\/section>\n<\/section>\n<\/blockquote>\n


<\/p>\n

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

\u80cc\u666f<\/strong><\/span><\/p>\n

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

 <\/span><\/p>\n

\u968f\u7740\u5de5\u4e1a\u5316\u4e0e\u57ce\u5e02\u5316\u8fdb\u7a0b\u7684\u52a0\u5feb\uff0c\u5927\u6c14\u6c61\u67d3\u6210\u4e3a\u5f53\u4eca\u4e16\u754c\u9762\u4e34\u7684\u6700\u4e25\u5cfb\u6311\u6218\u4e4b\u4e00\u3002\u636e\u4e16\u754c\u536b\u751f\u7ec4\u7ec7\u62a5\u9053\uff0c\u5168\u7403\u6bcf\u5e74\u7ea6\u6709700\u4e07\u4eba\u56e0\u5927\u6c14\u6c61\u67d3\u800c\u8fc7\u65e9\u6b7b\u4ea1[1]\u3002\u5927\u6c14\u6c61\u67d3\u7684\u6eaf\u6e90\u53ca\u9884\u62a5\u80fd\u591f\u6709\u6548\u5730\u6307\u5bfc\u4e2a\u4eba\u7684\u51fa\u884c\u53ca\u653f\u5e9c\u7684\u51b3\u7b56\uff0c\u5982\u5de5\u5382\u505c\u4ea7\u9650\u4ea7\u3001\u8f66\u8f86\u9650\u884c\u7b49\uff0c\u5177\u6709\u91cd\u8981\u7684\u610f\u4e49\u3002\u8fd1\u5e74\u6765\uff0c\u6211\u56fd\u5728\u5927\u6c14\u6c61\u67d3\u9632\u6cbb\u65b9\u9762\u6295\u5165\u4e86\u5927\u91cf\u7684\u4eba\u529b\u7269\u529b\uff0c\u4f46\u4e25\u91cd\u7684\u5927\u6c14\u6c61\u67d3\u4e8b\u4ef6\u4ecd\u7136\u9891\u7e41\u53d1\u751f\u3002\u7136\u800c\u5927\u6c14\u6c61\u67d3\u9884\u62a5\u95ee\u9898\u5177\u6709\u5f88\u5927\u7684\u6311\u6218\u6027\uff0c\u4e3b\u8981\u8868\u73b0\u5728\u4ee5\u4e0b\u51e0\u70b9\uff1a\u9996\u5148\uff0c\u5927\u6c14\u6c61\u67d3\u662f\u5178\u578b\u7684\u590d\u6742\u7cfb\u7edf[3]\uff0c\u6c61\u67d3\u7269\u4e4b\u95f4\u4f1a\u4ea7\u751f\u590d\u6742\u7684\u5316\u5b66\u53cd\u5e94\uff0c\u751f\u6210\u4e8c\u6b21\u6c61\u67d3\u7269\uff0c\u4f7f\u5f97\u6574\u4e2a\u7cfb\u7edf\u4ea7\u751f\u6574\u4f53\u5927\u4e8e\u90e8\u5206\u4e4b\u548c\u7684\u6548\u5e94\uff1b\u5728\u65f6\u95f4\u548c\u7a7a\u95f4\u7ef4\u5ea6\u5747\u5b58\u5728\u957f\u7a0b\u76f8\u5173\u6027[4]\uff1b\u6b64\u5916\uff0c\u878d\u5165\u6c14\u8c61\u4e0e\u73af\u5883\u4e13\u4e1a\u7684\u9886\u57df\u77e5\u8bc6\u5bf9\u8be5\u95ee\u9898\u7684\u7814\u7a76\u81f3\u5173\u91cd\u8981[5]\u3002<\/span><\/p>\n


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


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

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

\u76f8\u5173\u5de5\u4f5c<\/strong><\/span><\/p>\n

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

 <\/span><\/p>\n

\u73af\u5883\u9886\u57df\u7684\u4f20\u7edf\u7684\u7814\u7a76\u624b\u6bb5\u591a\u57fa\u4e8e\u8fd8\u539f\u8bba\u7684\u601d\u60f3\uff0c\u4f7f\u7528\u540d\u4e3a Community Multiscale Air Quality Modeling System <\/span>(CMAQ)<\/span> [6]\u7684\u7cfb\u7edf\u8fdb\u884c\u7814\u7a76\uff0c\u901a\u8fc7\u5fae\u5206\u65b9\u7a0b\u53bb\u7a77\u4e3e\u5176\u4e2d\u6d89\u53ca\u5230\u7684\u52a8\u529b\u5b66\u8fc7\u7a0b\u3002\u867d\u7136\u7cfb\u7edf\u6709\u5f88\u5f3a\u7684\u53ef\u89e3\u91ca\u6027\u548c\u4e13\u4e1a\u6027\uff0c\u4f46\u662f\u7531\u4e8e\u6a21\u578b\u8fd0\u884c\u901f\u5ea6\u592a\u6162\u4e14\u51c6\u786e\u7387\u8f83\u4f4e\uff0c\u53ea\u80fd\u7528\u4e8e\u5386\u53f2\u6570\u636e\u7684\u590d\u76d8\u5206\u6790\u4e2d\u3002<\/span><\/p>\n


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

\u800c\u57fa\u4e8e\u6df1\u5ea6\u5b66\u4e60\u7684\u65b9\u6cd5\u4f7f data-driven \u6570\u636e\u9a71\u52a8\u7684\u8303\u5f0f\uff0c\u53d7\u76ca\u4e8e\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u5f3a\u5927\u7684\u6570\u636e\u62df\u5408\u80fd\u529b\uff0c\u8fd9\u7c7b\u65b9\u6cd5\u5f80\u5f80\u5177\u6709\u5f88\u9ad8\u7684\u51c6\u786e\u7387\uff0c\u5e76\u4e14\u80fd\u591f\u8fdb\u884c\u5b9e\u65f6\u7684\u96fe\u973e\u9884\u62a5\uff0c\u4f46\u5f53\u524d\u65b9\u6cd5\u5728\u9886\u57df\u77e5\u8bc6\u7684\u878d\u5408\u4e0a\u8868\u73b0\u5f97\u4e0d\u597d\u3002\u8fd1\u5e74\u6765\uff0c\u56fe\u795e\u7ecf\u7f51\u7edc<\/span>\uff08GNN\uff09<\/span>[7]\u7684\u7814\u7a76\u6210\u4e3a\u6df1\u5ea6\u5b66\u4e60\u9886\u57df\u7684\u70ed\u70b9\u3002\u8fd9\u7c7b\u65b9\u6cd5\u57fa\u4e8e Graph \u7f51\u7edc\u7684\u5f62\u5f0f\u8fdb\u884c\u5efa\u6a21\uff0c\u5176\u4e2d\u8282\u70b9\u8868\u793a\u89c2\u6d4b\u7ad9\uff0c\u8fde\u8fb9\u8868\u793a\u7ad9\u70b9\u4e4b\u95f4\u7684\u7a7a\u95f4\u5173\u7cfb\u3002\u8fd9\u7c7b\u65b9\u6cd5\u4e0d\u50cf\u57fa\u4e8e CNN \u7684\u65b9\u6cd5\u4f1a\u5f15\u5165\u4e0d\u51c6\u786e\u7684\u5dee\u503c\u6570\u636e\uff0cGNN \u7684\u65b9\u6cd5\u5efa\u6a21\u66f4\u52a0\u81ea\u7136\uff0c\u4e14\u53ef\u4ee5\u878d\u5165\u590d\u6742\u7f51\u7edc\u7684\u5efa\u6a21\u65b9\u5f0f[3]\uff0c\u5438\u5f15\u4e86\u4f17\u591a\u7814\u7a76\u8005\u7684\u5173\u6ce8\u3002<\/span><\/p>\n


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

\u5317\u4eac\u5927\u5b66\u5730\u7403\u4e0e\u7a7a\u95f4\u79d1\u5b66\u5b66\u9662\u53d1\u8868\u4e8e\u300aScience of the Total Environment\u300b\u7684\u6a21\u578b GC-LSTM[8] \u662f\u4e4b\u524d\u7684 SOTA \u6a21\u578b\uff0c\u4f46\u8be5\u5de5\u4f5c\u7814\u7a76\u7684\u533a\u57df\u8f83\u5c0f<\/span>\uff08\u4eac\u6d25\u5180\uff09<\/span>\uff0c\u4e14\u53d7\u9650\u4e8e GCN [9]\u6846\u67b6\u7684\u7ea6\u675f\uff0c\u8fde\u8fb9\u65e0\u6cd5\u663e\u5f0f\u5730\u8f93\u5165\u98ce\u5bf9\u96fe\u973e\u4f20\u64ad\u7684\u5f71\u54cd\u3002<\/span><\/p>\n


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

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


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

PM2.5<\/sub>-GNN\u6a21\u578b<\/strong><\/span><\/p>\n

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

 <\/span><\/p>\n

\u672c\u6587\u5728\u5982\u4e0b\u56fe\u6240\u793a\u7684\u5730\u7406\u7a7a\u95f4\u8fdb\u884c\u7814\u7a76<\/span>\uff08103\u00b0E-122\u00b0E 28\u00b0N-42\u00b0N\uff09<\/span>\uff0c\u8be5\u533a\u57df\u5305\u62ec\u4e86\u6211\u56fd\u53d7\u96fe\u973e\u5f71\u54cd\u6700\u4e25\u91cd\u7684\u51e0\u4e2a\u57ce\u5e02\u7fa4\u3002\u5728\u6784\u5efa\u96fe\u973e\u4f20\u8f93\u7f51\u7edc\u7684\u65f6\u5019\uff0c\u4e0d\u4ec5\u9700\u8981\u8003\u8651\u4e24\u4e2a\u57ce\u5e02\u4e4b\u95f4\u8ddd\u79bb\uff0c\u8fd8\u8981\u8003\u8651\u4e2d\u95f4\u662f\u5426\u6709\u5f71\u54cd\u4f20\u8f93\u7684\u9ad8\u5c71<\/span>\uff08\u96fe\u973e\u4e00\u822c\u5206\u5e03\u57281200km\u4ee5\u4e0b[10]\uff09<\/span>\u3002<\/span><\/p>\n


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

<\/p>\n


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

\u6587\u7ae0\u7684\u57fa\u672c\u601d\u8def\u5982\u4e0b\uff1a\u5bf9\u6c61\u67d3\u7684\u6210\u56e0\u8fdb\u884c\u5206\u6790\u4e0e\u7b80\u5316\uff0c\u4e00\u4e2a\u57ce\u5e02\u6c61\u67d3\u7269\u6765\u6e90\u4e8e\u672c\u5730\u6392\u653e\u548c\u5916\u5730\u8f93\u9001\uff0c\u98ce\u662f\u4f20\u8f93\u7684\u4e3b\u8981\u52a8\u529b\uff0c\u5176\u4ed6\u6c14\u8c61\u56e0\u7d20\u6bd4\u5982\u8fb9\u754c\u5c42\u9ad8\u5ea6\u3001\u964d\u96e8\u3001\u6e7f\u5ea6\u7b49\u90fd\u4f1a\u5f71\u54cd\u6c61\u67d3\u7269\u5728\u672c\u5730\u7684\u7d2f\u79ef\u4e0e\u6d88\u6563\u3002\u57fa\u4e8e\u590d\u6742\u7f51\u7edc\u7684\u89c2\u70b9\uff0c\u5c06\u6c61\u67d3\u7269\u9897\u7c92\u770b\u6210\u662f\u5728\u6240\u6784\u5efa\u7684\u57ce\u5e02\u6c61\u67d3\u7269\u4f20\u8f93\u7f51\u7edc\u4e0a\u968f\u673a\u6e38\u8d70\u7684\u7c92\u5b50\u3002\u7528\u5faa\u73af\u795e\u7ecf\u7f51\u7edc\u53d8\u4f53GRU\u4e0e\u56fe\u795e\u7ecf\u7f51\u7edcGNN\u7684\u7ec4\u5408\u8fdb\u884c\u52a8\u529b\u5b66\u7684\u5efa\u6a21\uff0c\u5206\u522b\u5bf9\u5e94\u4e8e\u5927\u6c14\u6c61\u67d3\u7684\u52a8\u529b\u5b66\u7ec4\u6210\u90e8\u5206\u3002<\/span><\/p>\n


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

<\/p>\n


<\/p>\n

\u6211\u4eec\u5c06\u6c14\u8c61\u6570\u636e\u901a\u8fc7\u8282\u70b9\u5c5e\u6027\u3001\u8fde\u8fb9\u5c5e\u6027\u7684\u5f62\u5f0f\u8f93\u5165\u5230\u7f51\u7edc\u4e2d\u3002\u800c\u96fe\u973e\u7684\u6d53\u5ea6\u5b9a\u4e49\u4e3a\u8282\u70b9\u72b6\u6001\u3002\u90a3\u4e48\u672c\u6587\u6240\u89e3\u51b3\u7684\u95ee\u9898\u53ef\u4ee5\u88ab\u6784\u5efa\u4e3a\u8f93\u5165\u8d77\u59cb\u65f6\u523b\u7684\u96fe\u973e\u89c2\u6d4b\u503c\u4ee5\u53ca\u672a\u6765\u7684\u6c14\u8c61\u9884\u62a5\u6570\u636e<\/span>\uff08\u6570\u636e\u6765\u6e90\u4e8e GFS\uff0c\u53ef\u4ee5\u4f5c\u4e3a\u5df2\u77e5\u6570\u636e\uff09<\/span>\uff0c\u8f93\u51fa\u672a\u6765\u96fe\u973e\u9884\u62a5\u6570\u636e\u3002\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0c\u8fb9\u754c\u5c42\u9ad8\u5ea6\u662f\u5bf9\u96fe\u973e\u9884\u62a5\u8d77\u5230\u5173\u952e\u6027\u4f5c\u7528\u7684\u91cf\uff0c\u4f46\u662f\u5f80\u5f80\u4f1a\u88ab\u8ba1\u7b97\u673a\u9886\u57df\u7684\u6587\u7ae0\u6240\u5ffd\u7565\uff0c\u672c\u6587\u5c06\u5176\u4f5c\u4e3a\u8282\u70b9\u5c5e\u6027\u8f93\u5165\u8fdb\u6765\u3002<\/span><\/p>\n


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

\u4e0b\u56fe\u662f PM2.5<\/sub>-GNN \u7684\u6a21\u578b\u6846\u56fe\u3002\u57fa\u4e8e GNN \u7684\u6d88\u606f\u4f20\u9012\u4e0e\u90bb\u5c45\u6c47\u805a\u673a\u5236[7]\uff0c\u5bf9\u4e8e Graph \u4e2d\u7684\u6bcf\u4e2a\u8282\u70b9<\/span>\uff08\u57ce\u5e02\uff09<\/span>\u90fd\u8981\u4e0e\u5176\u5177\u6709\u8fde\u8fb9\u7684\u8282\u70b9\u8fdb\u884c\u4fe1\u606f\u7684\u4ea4\u4e92\u3002\u5176\u4e2d\u533a\u522b\u4e8e GC-LSTM \u7684\u4e3b\u8981\u662f\u8fde\u8fb9\u4e4b\u95f4\u7684\u6d88\u606f\u4f20\u9012\u673a\u5236\uff0c\u878d\u5408\u4e86\u4e0e PM2.5<\/sub> \u4f20\u64ad\u76f8\u5173\u7684\u8fde\u8fb9\u5c5e\u6027<\/span>\uff08\u98ce\u901f\u3001\u98ce\u5411\u3001\u8282\u70b9\u8ddd\u79bb\u3001\u8282\u70b9\u65b9\u5411\u7b49\uff09<\/span>\uff0c\u5e76\u4e14\u8fc7\u4e00\u4e2a MLP \u591a\u5c42\u611f\u77e5\u673a\uff0c\u7531\u6b64\u6765\u5efa\u6a21\u57ce\u5e02\u4e4b\u95f4\u5728\u98ce\u5f71\u54cd\u4e0b\u6240\u4ea7\u751f\u7684 PM2.5<\/sub> \u5728\u7a7a\u95f4\u4e0a\u7684\u975e\u7ebf\u6027\u4f9d\u8d56\u5173\u7cfb\u3002<\/span><\/p>\n


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

<\/p>\n


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

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

\u5b9e\u9a8c<\/strong><\/span><\/p>\n

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

 <\/span><\/p>\n

\u672c\u6587\u5728\u4e00\u4e2a\u56db\u5e74\u7684\u6570\u636e\u96c6\u4e0a\u505a\u5b9e\u9a8c<\/span>\uff08\u5212\u5206\u4e3a3\u4e2a\u5b50\u6570\u636e\u96c6\uff09<\/span>\u3002<\/span><\/p>\n


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

<\/p>\n


<\/p>\n

\u5728\u4e09\u4e2a\u5b50\u6570\u636e\u96c6\u4e0a\uff0c\u90fd\u8fdb\u884c3\u5929\u7684 PM2.5<\/sub> \u9884\u62a5\u5b9e\u9a8c\u3002\u4f7f\u7528 RMSE \u548c CSI[2] \u4e24\u4e2a\u6307\u6807\uff0c\u5176\u4e2d RMSE \u8d8a\u4f4e\u8d8a\u597d\uff0cCSI \u8d8a\u9ad8\u8d8a\u597d\u3002\u80fd\u770b\u51fa\u6765\uff0c\u5728\u6240\u6709\u5b50\u6570\u636e\u96c6\u4e0a\u5747\u80fd\u51fb\u8d25\u5bf9\u6bd4\u6a21\u578b GC-LSTM\u3002\u5e76\u4e14\uff0c\u5bf9\u6a21\u578b\u7684\u7ec4\u4ef6\u8fdb\u884c\u4e86\u6709\u6548\u6027\u7684\u9a8c\u8bc1\u3002\u53bb\u6389\u5176\u4e2d\u7684\u8fb9\u754c\u5c42\u9ad8\u5ea6 PBL\u6216\u8005 export \u7ec4\u4ef6\uff0c\u90fd\u4f1a\u4f7f\u5f97\u7ed3\u679c\u53d8\u5dee\u3002<\/span><\/p>\n


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

<\/p>\n


<\/p>\n

\u6211\u4eec\u9009\u53d6\u4e86\u897f\u5b89\u8fd9\u4e00\u6709\u4ee3\u8868\u6027\u7684\u57ce\u5e02\u505a\u6a21\u578b\u7684\u5bf9\u6bd4\u5b9e\u9a8c\uff0c\u5176\u4e2d\u84dd\u7ebf\u8868\u793a\u771f\u5b9e\u503c\uff0c\u9ec4\u7ebf\u8868\u793a\u6a21\u578b\u7684\u9884\u6d4b\uff0c\u80fd\u770b\u51fa\u6765\uff0cPM2.5<\/sub>-GNN \u6a21\u578b\u66f4\u52a0\u8d34\u5408\u771f\u5b9e\u503c\u3002<\/span><\/p>\n


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

<\/p>\n

\u6570\u636e\u548c\u4ee3\u7801\u5747\u5df2\u5f00\u6e90\uff1ahttps:\/\/github.com\/shawnwang-tech\/PM2.5<\/sub>-GNN<\/span><\/p>\n


<\/p>\n


<\/p>\n

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

\u603b\u7ed3<\/strong><\/span><\/p>\n

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

 <\/span><\/p>\n

\u672c\u6587\u57fa\u4e8e\u590d\u6742\u7f51\u7edc\u7684\u89c2\u70b9\uff0c\u5c06\u6c61\u67d3\u7269\u9897\u7c92\u770b\u6210\u662f\u5728\u6240\u6784\u5efa\u7684\u57ce\u5e02\u96fe\u973e\u4f20\u8f93\u7f51\u7edc\u4e0a\u968f\u673a\u6e38\u8d70\u7684\u7c92\u5b50\u3002\u4f7f\u7528 GRU \u4e0e GNN \u7684\u7ec4\u5408\u6765\u5b66\u4e60\u96fe\u973e\u7684\u52a8\u529b\u5b66\u7279\u5f81\u3002\u5176\u4e2d\uff0cGRU\u7684 cell \u5b58\u50a8\u548c\u9057\u5fd8\u95e8\u522b\u5efa\u6a21\u6c61\u67d3\u7269\u7684\u7d2f\u79ef\u4e0e\u7ad6\u76f4\u65b9\u5411\u4e0a\u7684\u6d88\u6563\uff0cGNN \u7684\u6d88\u606f\u4f20\u9012\u4e0e\u90bb\u5c45\u6c47\u805a\u5206\u5efa\u6a21\u6c61\u67d3\u7269\u5728\u98ce\u5f71\u54cd\u4e0b\u7684\u4f20\u64ad\u3002\u5728\u57fa\u4e8e Graph \u5efa\u6a21\u7684 PM2.5<\/sub> \u9884\u62a5\u4e2d\u8fbe\u5230\u4e86 state-of-the-art \u7684\u6548\u679c\u3002\u6b64\u5916\uff0c\u8be5\u6a21\u578b\u8fd8\u53ef\u4ee5\u88ab\u8fc1\u79fb\u5230\u5176\u4ed6\u65f6\u7a7a\u6570\u636e\u9884\u6d4b\u4e2d\uff0c\u6bd4\u5982\u81ed\u6c27\u9884\u62a5\u3001\u6c99\u5c18\u66b4\u9884\u62a5\u7b49\u3002<\/span><\/p>\n


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

<\/section>\n

\u96c6\u667a\u5b66\u56ed\u5206\u4eab\u89c6\u9891\u53ef\u89c1\uff1a<\/span><\/p>\n

https:\/\/campus.swarma.org\/course\/1996<\/span>\uff08\u6ce8\uff1a\u8bfe\u7a0b\u7b2c\u4e8c\u8282<\/span>\u662f\u5bf9\u672c\u8bba\u6587\u7684\u89e3\u8bfb\uff09<\/span><\/p>\n

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

\u6216\u53c2\u8003B\u7ad9\u89c6\u9891\uff1a<\/span>https:\/\/www.bilibili.com\/video\/BV1xV411y7Ri<\/span><\/p>\n


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


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

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

[1] \u7a7a\u6c14\u6c61\u67d3\u5982\u4f55\u635f\u5bb3\u6211\u4eec\u7684\u5065\u5eb7
https:\/\/www.who.int\/zh\/air-pollution\/news-and-events\/how-air-pollution-is-destroying-our-health<\/span><\/section>\n
[2] Shi, X. et al. 2015. Convolutional LSTM network: A machine learning approach for precipitation nowcasting. Advances in Neural Information Processing Systems (2015), 802\u2013810.<\/span><\/section>\n
[3] Song, C. et al. 2019. Modeling Air Pollution Transmission Behavior as Complex Network and Mining Key Monitoring Station. IEEE Access. 7, (2019), 121245\u2013121254. DOI:https:\/\/doi.org\/10.1109\/access.2019.2936613. <\/span><\/section>\n
[4] Hu, J. et al. 2014. Spatial and temporal variability of PM 2.5 and PM 10 over the North China Plain and the Yangtze River Delta, China. Atmospheric Environment. 95, (2014), 598\u2013609.
DOI:https:\/\/doi.org\/10.1016\/j.atmosenv.2014.07.019. <\/span><\/section>\n
[5] Chen, Z. et al. 2020. Influence of meteorological conditions on PM2.5<\/sub> concentrations across China: A review of methodology and mechanism. Environment International. 139, (Jun. 2020), 105558. DOI:https:\/\/doi.org\/10.1016\/j.envint.2020.105558<\/span><\/section>\n
[6] https:\/\/www.epa.gov\/cmaq<\/span><\/section>\n
[7] Battaglia, P.W. et al. 2018. Relational inductive biases, deep learning, and graph networks. (2018).<\/span><\/section>\n
[8] Qi, Y. et al. 2019. A hybrid model for spatiotemporal forecasting of PM 2.5 based on graph convolutional neural network and long short-term memory. Science of the Total Environment. 664, February (2019), 1\u201310. DOI:https:\/\/doi.org\/10.1016\/j.scitotenv.2019.01.333. <\/span><\/section>\n
[9] Kipf, T.N. and Welling, M. 2017. Semi-supervised classification with graph convolutional networks. 5th International Conference on Learning Representations, ICLR 2017 – Conference Track Proceedings (Sep. 2017).<\/span><\/section>\n
[10] I. Uno, N. Sugimoto, A. Shimizu, K. Yumimoto, Y. Hara, and Z. Wang, \u201cRecord heavy PM2.5<\/sub> air pollution over China in January 2013: Vertical and horizontal dimensions,\u201d Sci. Online Lett. Atmos., vol. 10, no. 1, pp. 136\u2013140, 2014.<\/span><\/section>\n
 <\/span><\/section>\n
\n
\n
\n
\u4f5c\u8005<\/span>\uff1a\u738b\u7855<\/span><\/span><\/section>\n
\u7f16\u8f91\uff1a\u9093\u4e00\u96ea<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n
\n
\n

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

\u590d\u6742\u79d1\u5b66\u6700\u65b0\u8bba\u6587<\/strong><\/span><\/p>\n

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


<\/section>\n

\u96c6\u667a\u6591\u56fe\u9876\u520a\u8bba\u6587\u901f\u9012\u680f\u76ee\u4e0a\u7ebf\u4ee5\u6765\uff0c\u6301\u7eed\u6536\u5f55\u6765\u81eaNature\u3001Science<\/em>\u7b49\u9876\u520a\u7684\u6700\u65b0\u8bba\u6587\uff0c\u8ffd\u8e2a\u590d\u6742\u7cfb\u7edf\u3001\u7f51\u7edc\u79d1\u5b66\u3001\u8ba1\u7b97\u793e\u4f1a\u79d1\u5b66\u7b49\u9886\u57df\u7684\u524d\u6cbf\u8fdb\u5c55\u3002\u73b0\u5728\u6b63\u5f0f\u63a8\u51fa\u8ba2\u9605\u529f\u80fd\uff0c\u6bcf\u5468\u901a\u8fc7\u5fae\u4fe1\u670d\u52a1\u53f7\u300c\u96c6\u667a\u6591\u56fe\u300d\u63a8\u9001\u8bba\u6587\u4fe1\u606f\u3002\u626b\u63cf\u4e0b\u65b9\u4e8c\u7ef4\u7801\u5373\u53ef\u4e00\u952e\u8ba2\u9605\uff1a<\/span><\/p>\n


<\/p>\n

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


<\/p>\n

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

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


<\/section>\n
\u751f\u547d\u590d\u6742\u6027\u7cfb\u5217\u8bfb\u4e66\u4f1a<\/strong><\/a>
<\/section>\n
\u56fe\u795e\u7ecf\u7f51\u7edc\u524d\u6cbf\u7efc\u8ff0\uff1a\u52a8\u6001\u56fe\u7f51\u7edc<\/span><\/strong><\/a>
<\/section>\n
\u5982\u4f55\u7528\u56fe\u795e\u7ecf\u7f51\u7edc\u505a\u4e00\u4e2a\u7269\u7406\u5f15\u64ce | \u8bba\u6587\u7b14\u8bb0<\/span><\/strong><\/a>
<\/section>\n
\u590d\u6742\u7f51\u7edc\u52a8\u529b\u5b66\u673a\u5668\u5b66\u4e60\u81ea\u52a8\u5efa\u6a21<\/span><\/strong><\/a>
<\/section>\n
\u590d\u6742\u7cfb\u7edf\u81ea\u52a8\u5efa\u6a21\u7efc\u8ff0\uff1a\u63cf\u8ff0\u3001\u9884\u6d4b\u4e0e\u7406\u8bba\u53d1\u73b0<\/span><\/strong><\/a><\/section>\n
\u52a0\u5165\u96c6\u667a\uff0c\u4e00\u8d77\u590d\u6742\uff01<\/a><\/span><\/strong><\/span><\/section>\n

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


<\/p>\n

\n


<\/p>\n


<\/p>\n

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

\u96c6\u667a\u4ff1\u4e50\u90e8QQ\u7fa4\uff5c877391004<\/span><\/strong><\/p>\n

\u5546\u52a1\u5408\u4f5c\u53ca\u6295\u7a3f\u8f6c\u8f7d\uff5cswarma@swarma.org
<\/span><\/strong><\/p>\n

\n

\u25c6 <\/span>\u25c6 <\/span>\u25c6<\/span><\/span><\/strong><\/strong><\/h1>\n<\/section>\n

\u641c\u7d22\u516c\u4f17\u53f7\uff1a\u96c6\u667a\u4ff1\u4e50\u90e8<\/span><\/strong><\/span><\/p>\n


<\/p>\n

\u52a0\u5165\u201c\u6ca1\u6709\u56f4\u5899\u7684\u7814\u7a76\u6240\u201d<\/span><\/strong><\/span><\/p>\n

\n

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

\u8ba9\u82f9\u679c\u7838\u5f97\u66f4\u731b\u70c8\u4e9b\u5427\uff01<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n


<\/p>\n


<\/p>\n

\ud83d\udc47\u70b9\u51fb\u201c\u9605\u8bfb\u539f\u6587\u201d\uff0c\u8ffd\u8e2a\u590d\u6742\u79d1\u5b66\u9876\u520a\u8bba\u6587<\/span><\/span><\/p>\n<\/p><\/div>\n","protected":false},"excerpt":{"rendered":"

\u5bfc\u8bed \u672c\u6587\u603b\u7ed3\u4e86\u5317\u4eac\u5e08\u8303\u5927\u5b66\u7cfb\u7edf\u79d1\u5b66\u5b66\u9662\u5728\u8bfb\u535a\u58eb\u751f\u738b\u7855\u5728\u96c6\u667a-\u51ef\u98ce\u7814\u8bfb\u8425\u4e0a\u7684\u5206\u4eab\uff0c\u4ecb\u7ecd\u5176\u53d1\u8868\u4e8e SIGSPATIAL 2020 \u7684\u5de5\u4f5c\u300aPM2.5-GNN: A Domain Knowledge Enhanced Graph Neural Network For PM2.5 Forecasting\u300b…<\/p>\n","protected":false},"author":1,"featured_media":21916,"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\/21925"}],"collection":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=21925"}],"version-history":[{"count":0,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/21925\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/media\/21916"}],"wp:attachment":[{"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=21925"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=21925"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=21925"},{"taxonomy":"special","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fspecial&post=21925"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}