{"id":17285,"date":"2019-10-17T17:56:56","date_gmt":"2019-10-18T09:50:11","guid":{"rendered":"https:\/\/swarma.org\/?p=17285"},"modified":"2019-10-18T20:12:13","modified_gmt":"2019-10-18T12:12:13","slug":"4c4c485eeb","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=17285","title":{"rendered":"\u590d\u6742\u7cfb\u7edf\u4e0e\u793e\u4f1a\u8ba1\u7b97\u7814\u8ba8\u4f1a\u542f\u52a8\u62a5\u540d | \u5317\u4eac\u8ba1\u7b97\u79d1\u5b66\u7814\u7a76\u4e2d\u5fc310\u670827\u65e5"},"content":{"rendered":"
\n
\"\u96c6\u667a\"<\/section>\n
\n
\n
\u5bfc\u8bed<\/strong><\/section>\n
\n
\n
\n

\u5317\u4eac\u8ba1\u7b97\u79d1\u5b66\u7814\u7a76\u4e2d\u5fc3\u5c06\u4e8e10\u670827\u65e5\uff08\u5468\u65e5\uff09\u5728\u5317\u4eac\u4e3e\u529e\u5168\u5929\u7684\u590d\u6742\u7cfb\u7edf\u4e0e\u793e\u4f1a\u8ba1\u7b97\u7814\u8ba8\u4f1a\u6d3b\u52a8\uff0c\u672c\u6b21\u6d3b\u52a8\u7531\u4e2d\u6587\u4fe1\u606f\u5b66\u4f1a\u793e\u4f1a\u5a92\u4f53\u5904\u7406\u4e13\u59d4\u4f1a\u793e\u4f1a\u7f51\u7edc\u5206\u6790\u4e13\u4e1a\u7ec4\uff08\u7b79\uff09\u8054\u5408\u7ec4\u7ec7\u3002\u4f1a\u8bae\u8bae\u7a0b\u4f9b\u5404\u4f4d\u540c\u597d\u53c2\u8003\u3002\u6b22\u8fce\u611f\u5174\u8da3\u7684\u670b\u53cb\u62a5\u540d\u53c2\u4e0e\u3002\u96c6\u667a\u4ff1\u4e50\u90e8\u5c06\u8fdb\u884c\u5168\u7a0b\u76f4\u64ad\u3002<\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\n

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

\u4f5c\u4e3a\u4e00\u4e2a\u8de8\u5b66\u79d1\u9886\u57df\uff0c\u590d\u6742\u7cfb\u7edf\u79d1\u5b66\u5438\u6536\u4e86\u8bb8\u591a\u4e0d\u540c\u9886\u57df\u7684\u8d21\u732e\uff0c\u6db5\u76d6\u4e86\u8bb8\u591a\u4e0d\u540c\u5b66\u79d1\u95ee\u9898\u7684\u7814\u7a76\u65b9\u6cd5\uff0c\u5305\u62ec\u7edf\u8ba1\u7269\u7406\u5b66\u3001\u4fe1\u606f\u8bba\u3001\u8ba1\u7b97\u673a\u79d1\u5b66\u3001\u751f\u7269\u5b66\u7b49\uff0c\u8fd1\u5e74\u6765\u53d6\u5f97\u4e30\u7855\u7684\u7814\u7a76\u6210\u679c\u3002<\/span><\/p>\n

\u540c\u6837\u4f5c\u4e3a\u4e00\u4e2a\u8de8\u5b66\u79d1\u7684\u65b0\u5174\u7814\u7a76\u9886\u57df\uff0c\u793e\u4f1a\u8ba1\u7b97\u8fd1\u5e74\u6765\u84ec\u52c3\u53d1\u5c55\uff0c\u5e76\u5f97\u5230\u56fd\u5185\u5916\u4fe1\u606f\u79d1\u5b66\u53ca\u76f8\u5173\u4ea4\u53c9\u5b66\u79d1\u9886\u57df\u7684\u9ad8\u5ea6\u91cd\u89c6\uff0c\u6b63\u5728\u5b9e\u73b0\u793e\u4f1a\u79d1\u5b66\u3001\u4fe1\u606f\u79d1\u5b66\u548c\u7ba1\u7406\u79d1\u5b66\u591a\u5b66\u79d1\u4ea4\u53c9\u7814\u7a76\u7684\u5b9e\u8d28\u8fdb\u5c55\u548c\u878d\u5408\u3002<\/span><\/p>\n

\u4e3a\u4e86\u4fc3\u8fdb\u8de8\u5b66\u79d1\u65b0\u5174\u9886\u57df\u7684\u53d1\u5c55\uff0c\u5317\u4eac\u8ba1\u7b97\u79d1\u5b66\u7814\u7a76\u4e2d\u5fc3<\/strong>\u5c06\u4e8e10\u670827\u65e5\uff08\u5468\u65e5\uff09\u5728\u5317\u4eac\u4e3e\u529e\u5168\u5929\u7684\u590d\u6742\u7cfb\u7edf\u4e0e\u793e\u4f1a\u8ba1\u7b97\u7814\u8ba8\u4f1a\u6d3b\u52a8\u3002\u672c\u6b21\u6d3b\u52a8\u7531\u4e2d\u6587\u4fe1\u606f\u5b66\u4f1a\u793e\u4f1a\u5a92\u4f53\u5904\u7406\u4e13\u59d4\u4f1a\u793e\u4f1a\u7f51\u7edc\u5206\u6790\u4e13\u4e1a\u7ec4\uff08\u7b79\uff09<\/strong>\u8054\u5408\u7ec4\u7ec7\uff0c\u6b22\u8fce\u611f\u5174\u8da3\u7684\u670b\u53cb\u62a5\u540d\u53c2\u52a0\u5f53\u5929\u6d3b\u52a8\u3002<\/span><\/p>\n

\u96c6\u667a\u4ff1\u4e50\u90e8\u5c06\u5bf9\u7814\u8ba8\u4f1a\u4e0a\u591a\u4f4d\u5b66\u8005\u7684\u4e3b\u9898\u5206\u4eab\u8fdb\u884c\u5168\u7a0b\u89c6\u9891\u76f4\u64ad\u3002<\/span><\/p>\n

\u4f1a\u8bae\u4fe1\u606f<\/strong><\/p>\n

\u65f6\u95f4\uff1a<\/span><\/strong>2019\u5e7410\u670827\u65e5<\/span><\/section>\n
\u4e3b\u529e\u65b9\uff1a<\/span><\/strong>\u5317\u4eac\u8ba1\u7b97\u79d1\u5b66\u7814\u7a76\u4e2d\u5fc3<\/span><\/section>\n
\u8054\u5408\u4e3b\u529e\uff1a<\/span><\/strong>\u4e2d\u56fd\u4e2d\u6587\u4fe1\u606f\u5b66\u4f1a\u793e\u4f1a\u5a92\u4f53\u5904\u7406\u4e13\u59d4\u4f1a\u793e\u4ea4\u7f51\u7edc\u5206\u6790\u4e13\u4e1a\u7ec4\uff08\u7b79\uff09<\/span><\/section>\n
\u4f1a\u8bae\u5730\u5740\uff1a<\/span><\/strong>\u5317\u4eac\u5e02\u6d77\u6dc0\u533a\u897f\u5317\u65fa\u4e1c\u8def10\u53f7\u9662\u4e2d\u5173\u6751\u8f6f\u4ef6\u56ed\u4e8c\u671f\u4e1c\u533a9\u53f7\u697c\uff08\u641c\u7d22\u201c\u5317\u4eac\u8ba1\u7b97\u79d1\u5b66\u7814\u7a76\u4e2d\u5fc3\u201d\u5373\u53ef\uff09<\/span><\/section>\n
<\/section>\n
\u53ec\u96c6\u4eba\uff1a<\/span><\/strong>\u5415\u7433\u5a9b \u6c64\u96f7\u7ff0<\/section>\n
\u6ce8\u518c\u8d39\uff1a<\/span><\/strong>\u514d\u8d39<\/span><\/section>\n
\u53c2\u4f1a\u9700\u8981\u63d0\u524d\u62a5\u540d\uff0c\u62a5\u540d\u8005\u63d0\u4f9b\u4f1a\u8bae\u8d44\u6599\u888b\u548c\u5348\u9910<\/span><\/section>\n
\u62a5\u540d\u65b9\u5f0f\uff1a\u626b\u7801\u586b\u8868\u62a5\u540d\uff0c\u62a5\u540d\u6210\u529f\u4e4b\u540e\u5c06\u77ed\u4fe1\/\u90ae\u4ef6\u901a\u77e5<\/span><\/strong><\/section>\n
<\/section>\n
\u626b\u7801\u586b\u5199\u62a5\u540d\u4fe1\u606f<\/span><\/strong><\/section>\n

\u5468\u8fb9\u4f4f\u5bbf\uff1a<\/span><\/strong>\u4f1a\u8bae\u5468\u8fb9\u6709\u591a\u5bb6\u9152\u5e97\u65b9\u4fbf\u4f4f\u5bbf\uff0c\u8be6\u60c5\u89c1\u6587\u672b<\/span><\/p>\n

\u4f1a\u8bae\u65e5\u7a0b<\/strong><\/p>\n

\"\u96c6\u667a\"<\/section>\n
\n
\n
\n
\n
\n
\n
\n
\n

\u5609\u5bbe\u53ca\u62a5\u544a\u7b80\u4ecb<\/span><\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u5510\u6770 \u6e05\u534e\u5927\u5b66<\/strong><\/p>\n

\u9898\u76ee\uff1a<\/span><\/strong><\/section>\n
Graph Neural Networks and Applications<\/span><\/section>\n
\u6458\u8981\uff1a<\/span><\/strong><\/section>\n
Graph Neural networks (GNNs) and their variants have generalized deep learning methods into non-Euclidean graph data, bringing a substantial improvement on many graph mining tasks. In this talk, I will revisit graph convolutional networks and investigate how to improve their representation capacity. We discover that the performance of GNNs can be significantly improved with several simple and elegant refinements on the neighborhood aggregation and network sampling steps. Importantly, we show that some of the most expressive GNNs, e.g., the graph attention network, can be reformulated as a particular instance of our models. Extensive experiments on different types of graph benchmarks show that our proposed framework can significantly and consistently improve the graph classification accuracy when compared to state-of-the-art baselines.<\/span><\/section>\n
\"\u96c6\u667a\"<\/section>\n
\u7b80\u4ecb\uff1a<\/span><\/strong><\/section>\n
\u5510\u6770\uff0c\u6e05\u534e\u5927\u5b66\u8ba1\u7b97\u673a\u79d1\u5b66\u4e0e\u6280\u672f\u7cfb\u957f\u8058\u6559\u6388\uff0c\u8ba1\u7b97\u673a\u7cfb\u526f\u4e3b\u4efb\u3001\u6e05\u534e-\u5de5\u7a0b\u9662\u77e5\u8bc6\u667a\u80fd\u8054\u5408\u5b9e\u9a8c\u5ba4\u4e3b\u4efb\uff0c\u83b7\u6770\u9752\u3002\u7814\u7a76\u5174\u8da3\u5305\u62ec\uff1a\u793e\u4f1a\u7f51\u7edc\u5206\u6790\u3001\u6570\u636e\u6316\u6398\u3001\u673a\u5668\u5b66\u4e60\u548c\u77e5\u8bc6\u56fe\u8c31\u3002\u53d1\u8868\u8bba\u6587200\u4f59\u7bc7\uff0c\u62e5\u6709\u4e13\u522920\u4f59\u9879\u3002\u4e3b\u6301\u7814\u53d1\u4e86\u7814\u7a76\u8005\u793e\u4f1a\u7f51\u7edc\u6316\u6398\u7cfb\u7edfAMiner\uff0c\u5438\u5f15\u4e86220\u4e2a\u56fd\u5bb6\/\u5730\u533a1000\u591a\u4e07\u72ec\u7acbIP\u8bbf\u95ee\u3002\u66fe\u62c5\u4efb\u56fd\u9645\u671f\u520aACM TKDD\u7684\u6267\u884c\u4e3b\u7f16\u548c\u56fd\u9645\u4f1a\u8baeCIKM\u201916\u3001WSDM\u201915\u7684\u7a0b\u5e8f\u59d4\u5458\u4f1a\u4e3b\u5e2d\u3001KDD\u201918\u5927\u4f1a\u526f\u4e3b\u5e2d\u4ee5\u53caIEEE TKDE\u3001ACM TIST\u3001IEEE TBD\u7b49\u671f\u520a\u7f16\u59d4\u3002\u83b7\u82f1\u56fd\u7687\u5bb6\u5b66\u4f1a-\u725b\u987f\u9ad8\u7ea7\u5956\u5b66\u91d1\u3001CCF\u9752\u5e74\u79d1\u5b66\u5bb6\u5956\u3001\u56fd\u5bb6\u81ea\u7136\u79d1\u5b66\u57fa\u91d1\u59d4\u5458\u4f1a\u6770\u51fa\u9752\u5e74\u5b66\u8005\u3001\u5317\u4eac\u5e02\u79d1\u6280\u8fdb\u6b65\u4e00\u7b49\u5956\u3001\u4e2d\u56fd\u4eba\u5de5\u667a\u80fd\u5b66\u4f1a\u79d1\u6280\u8fdb\u6b65\u4e00\u7b49\u5956\u3001KDD’18 \u6770\u51fa\u8d21\u732e\u5956\u30022012\u5e74\u56fd\u5bb6\u4f18\u79c0\u9752\u5e74\u79d1\u5b66\u57fa\u91d1\u83b7\u5f97\u8005\u3002<\/span><\/section>\n

\u95eb\u5c0f\u52c7 \u5317\u4eac\u4ea4\u901a\u5927\u5b66<\/strong><\/p>\n

\u9898\u76ee\uff1a<\/span><\/strong><\/section>\n
\u793e\u4f1a\u5f15\u529b\u5b9a\u5f8b\u8ffd\u6839\u6eaf\u6e90<\/span><\/section>\n
\u6458\u8981\uff1a<\/span><\/strong><\/section>\n
\u5f15\u529b\u5b9a\u5f8b\u539f\u672c\u662f\u89e3\u91ca\u548c\u9884\u6d4b\u7269\u4f53\u4e4b\u95f4\u5f15\u529b\u4ea4\u4e92\u7684\u4e00\u4e2a\u57fa\u672c\u7269\u7406\u5b9a\u5f8b\uff0c\u4f46\u6709\u8da3\u7684\u662f\uff0c\u5728\u4eba\u53e3\u8fc1\u79fb\u3001\u4ea4\u901a\u51fa\u884c\u3001\u5546\u54c1\u8d38\u6613\u3001\u4fe1\u606f\u901a\u8baf\u3001\u793e\u4f1a\u4ea4\u5f80\u3001\u79d1\u7814\u5408\u4f5c\u7b49\u5927\u91cf\u4eba\u3001\u7269\u3001\u4fe1\u606f\u7684\u7a7a\u95f4\u4ea4\u4e92\u73b0\u8c61\u4e2d\uff0c\u90fd\u5b58\u5728\u7c7b\u4f3c\u4e07\u6709\u5f15\u529b\u5b9a\u5f8b\u7684\u89c4\u5f8b\u3002\u8fd9\u4e9b\u5e7f\u6cdb\u5b58\u5728\u7684\u793e\u4f1a\u5f15\u529b\u5b9a\u5f8b\u53cd\u6620\u7684\u90fd\u662f\u4e24\u5730\u4e4b\u95f4\u7684\u67d0\u79cd\u6d41\u91cf\u4e0e\u4e24\u5730\u89c4\u6a21\uff08\u5982\u4eba\u53e3\u3001GDP\u7b49\uff09\u7684\u4e58\u79ef\u6210\u6b63\u6bd4\u3001\u4e0e\u4e24\u5730\u8ddd\u79bb\u7684\u5e42\u5f8b\u6210\u53cd\u6bd4\u7684\u7279\u5f81\u3002\u4f46\u590d\u6742\u7684\u793e\u4f1a\u7cfb\u7edf\u4e2d\u4e3a\u4ec0\u4e48\u4f1a\u6709\u8fd9\u6837\u7b80\u5355\u7684\u5f15\u529b\u5b9a\u5f8b\u5b58\u5728\uff1f\u8fd9\u662f\u4e2a\u975e\u5e38\u6709\u8da3\u4e5f\u6709\u4ef7\u503c\u7684\u95ee\u9898\u3002\u672c\u62a5\u544a\u5c06\u4ece\u5927\u91cf\u793e\u4f1a\u5f15\u529b\u5b9a\u5f8b\u5b9e\u8bc1\u89c2\u5bdf\u51fa\u53d1\uff0c\u8ffd\u6839\u6eaf\u6e90\uff0c\u4ece\u6700\u5927\u71b5\u539f\u7406\u3001\u968f\u673a\u6548\u7528\u7406\u8bba\u548c\u535a\u5f08\u8bba\u7b49\u4e0d\u540c\u89c6\u89d2\u63a2\u7d22\u793e\u4f1a\u5f15\u529b\u5b9a\u5f8b\u80cc\u540e\u7684\u673a\u7406\u3002<\/span><\/section>\n
\"\u96c6\u667a\"<\/span><\/section>\n
\u7b80\u4ecb\uff1a<\/span><\/strong><\/section>\n
\u95eb\u5c0f\u52c7\uff0c\u7537\uff0c1980\u5e749\u6708\u51fa\u751f\uff0c\u535a\u58eb\uff0c\u5317\u4eac\u4ea4\u901a\u5927\u5b66\u4ea4\u901a\u7cfb\u7edf\u79d1\u5b66\u4e0e\u5de5\u7a0b\u7814\u7a76\u9662\u6559\u6388\u3002\u4e3b\u8981\u4ece\u4e8b\u590d\u6742\u7f51\u7edc\u548c\u4eba\u7c7b\u79fb\u52a8\u884c\u4e3a\u590d\u6742\u6027\u7814\u7a76\uff0c\u4e3b\u6301\u5305\u62ec\u4f18\u79c0\u9752\u5e74\u57fa\u91d1\u9879\u76ee\u5728\u5185\u7684\u56fd\u5bb6\u81ea\u7136\u79d1\u5b66\u57fa\u91d1\u9879\u76ee3\u9879\u3002\u5728Nature Communications\u3001Journal of The Royal Society Interface\u7b49\u671f\u520a\u4e0a\u53d1\u8868SCI\u68c0\u7d22\u8bba\u658730\u4f59\u7bc7\uff0c\u7814\u7a76\u6210\u679c\u88ab\u4e2d\u56fd\u79d1\u5b66\u62a5\u3001\u4e2d\u56fd\u6559\u80b2\u62a5\u3001BBC\u3001Nature\u3001AAAS EurekAlert!\u3001Science Daily\u3001Phys.org\u7b49\u56fd\u5185\u5916\u79d1\u6280\u5a92\u4f53\u5e7f\u6cdb\u62a5\u9053\u3002\u5176\u4e2d\u53d1\u8868\u5728Nature Communications\u7684\u7814\u7a76\u6210\u679c\u5f97\u5230\u4e86\u56fd\u5bb6\u9886\u5bfc\u4eba\u7684\u91cd\u8981\u6279\u793a\uff0c\u5e76\u5df2\u5728\u5317\u4eac\u7b49\u57ce\u5e02\u83b7\u5f97\u4e86\u63a8\u5e7f\u5e94\u7528\u3002<\/span><\/section>\n

\u80e1\u5ef6\u5e86 \u4e2d\u5c71\u5927\u5b66<\/strong><\/p>\n

\u9898\u76ee\uff1a<\/span><\/strong><\/section>\n
Revealing the Predictability of Intrinsic Structure in Complex Networks<\/span><\/section>\n
\u6458\u8981\uff1a<\/span><\/strong><\/section>\n
Structure prediction in networks is among the most important and widely studied problems in network science and machine learning, and has enormous contribution to various fields from biology, recommendation systems to social media. Despite the significant progress in prediction algorithms, the fundamental predictability of structures remains unclear as networks\u2019 complex underlying formation dynamics are usually unobserved or difficult to describe. Hence to date there has been a lack of theoretical guidance on the practical development of algorithms for their absolute performances. Here, for the first time, we find that the shortest compression length of a network structure can directly assess the structure predictability, with a universal linear relationship between the two quantities found in a broad range of real networks, hinting at a possible universal class among empirical networks. Specifically, shorter binary string length from compression leads to higher link predictability. We also analytically derive the physical origin of this linear relationship in artificial random networks. In addition, our finding leads to analytical results on maximum prediction accuracy adjustable for different requirements, and allows the estimation of the network dataset potential values through the size of the compressed network data file.<\/span><\/section>\n
\"\u96c6\u667a\"<\/span><\/section>\n
\u7b80\u4ecb\uff1a<\/span><\/strong><\/section>\n
\u80e1\u5ef6\u5e86\uff0c\u535a\u58eb\uff0c\u73b0\u5728\u4e3a\u4e2d\u5c71\u5927\u5b66\u6570\u636e\u4e0e\u8ba1\u7b97\u673a\u5b66\u9662\u526f\u6559\u6388\uff08\u4e2d\u5927\u767e\u4eba\u8ba1\u5212\uff09\uff0c\u535a\u58eb\u751f\u5bfc\u5e08\u30022011\u5e74\u6bd5\u4e1a\u4e8e\u5317\u4eac\u5e08\u8303\u5927\u5b66\u7cfb\u7edf\u79d1\u5b66\u5b66\u9662\uff0c\u83b7\u5f97\u7cfb\u7edf\u7406\u8bba\u65b9\u5411\u7406\u5b66\u535a\u58eb\u5b66\u4f4d\uff0c\u5e76\u83b7\u5f97\u5317\u4eac\u5e02\u4f18\u79c0\u535a\u58eb\u8bba\u6587\u5956\uff1b2011-2013\u5e74\u7ebd\u7ea6\u57ce\u5e02\u5927\u5b66Levich Institute \u535a\u58eb\u540e\u3002\u8fd1\u51e0\u5e74\u4e3b\u8981\u4ece\u4e8b\u5177\u6709\u56fe\u6216\u8005\u7f51\u7edc\u7ed3\u6784\u7684\u5927\u6570\u636e\u6316\u6398\u4e0e\u4eba\u5de5\u667a\u80fd\u7b97\u6cd5\u4e0e\u7406\u8bba\u7814\u7a76\u5de5\u4f5c\uff0c\u63a2\u7d22\u6570\u636e\u80cc\u540e\u7684\u81ea\u7136\u7269\u7406\u89c4\u5f8b\u3002\u53d1\u8868\u8bba\u6587 40 \u4f59\u7bc7\uff0c\u5176\u4e2d\u901a\u8baf\u3001\u7b2c\u4e00\u4f5c\u8005\u8bba\u6587 25 \u7bc7, \u5305\u62ec Nature Physics, PRL, PRX \u5404 1 \u7bc7\uff0cPNAS 2 \u7bc7\uff0cPRE 11 \u7bc7\uff0c\u5176\u4e2d Nature Physics \u8bba\u6587\u88ab\u9009\u5165\u8be5\u671f\u5c01\u9762\u63a8\u8350\u8bba\u6587.<\/span><\/section>\n

\u674e\u5927\u5e86 \u5317\u4eac\u822a\u7a7a\u822a\u5929\u5927\u5b66<\/strong><\/p>\n

\u9898\u76ee\uff1a<\/span><\/strong><\/section>\n
\u57ce\u5e02\u4ea4\u901a\u7684\u5065\u5eb7\u7ba1\u7406<\/span><\/section>\n
\u6458\u8981\uff1a<\/span><\/strong><\/section>\n
\u5982\u679c\u628a\u57ce\u5e02\u770b\u6210\u4e00\u4e2a\u751f\u547d\u4f53\uff0c\u4ea4\u901a\u7f51\u7edc\u5c31\u662f\u5176\u751f\u547d\u7ebf\u3002\u4ea4\u901a\u7f51\u7edc\u7684\u5065\u5eb7\u8fd0\u884c\u662f\u4ea4\u901a\u53ef\u9760\u6027\u7ba1\u7406\u7684\u6838\u5fc3\u95ee\u9898\u3002\u540c\u65f6\uff0c\u57ce\u5e02\u4ea4\u901a\u76f8\u5173\u7684\u4f4d\u7f6e\u670d\u52a1\u662f\u5178\u578b\u7684\u5927\u6570\u636e\u666f\uff0c\u76ee\u524d\u7684\u4f4d\u7f6e\u5927\u6570\u636e\u8ba1\u7b97\u4f9d\u7136\u9762\u4e34\u7740\u6570\u636e\u7f3a\u5931\u3001\u62bd\u6837\u504f\u5dee\u3001\u8ba1\u7b97\u590d\u6742\u7b49\u74f6\u9888\u95ee\u9898\uff0c\u5bfc\u81f4\u8f83\u96be\u51fa\u73b0\u53ef\u652f\u6301\u51b3\u7b56\u7684\u57ce\u5e02\u4ea4\u901a\u8bc4\u4f30\u3001\u9884\u6d4b\u548c\u8c03\u63a7\u65b9\u6cd5\uff0c\u6210\u4e3a\u76ee\u524d\u667a\u6167\u57ce\u5e02\u5b9e\u65bd\u7684\u4e3b\u8981\u56f0\u96be\u4e4b\u4e00\u3002\u9762\u5bf9\u8fd9\u4e9b\u6311\u6218\uff0c\u5c31\u9700\u8981\u57fa\u4e8e\u73b0\u6709\u79ef\u7d2f\u7684\u6d77\u91cf\u6570\u636e\uff0c\u6316\u6398\u76f8\u5173\u7684\u7cfb\u7edf\u8bed\u4e49\u7279\u5f81\u548c\u6f14\u5316\u884c\u4e3a\uff0c\u652f\u6491\u66f4\u6df1\u5c42\u6b21\u7684\u7b97\u6cd5\u8bbe\u8ba1\u3002\u8fd1\u5e74\u6765\u6211\u4eec\u57fa\u4e8e\u6e17\u6d41\u7406\u8bba\uff0c\u7ed3\u5408\u3002\u4ea4\u901a\u8def\u51b5\u6570\u636e\uff0c\u5bf9\u4ea4\u901a\u62e5\u5835\u4ece\u4ea7\u751f\u3001\u6f14\u5316\u5230\u6062\u590d\u7684\u5168\u5bff\u547d\u5468\u671f\u8fdb\u884c\u5206\u6790\uff0c\u6316\u6398\u7cfb\u7edf\u7684\u5f39\u6027\u89c4\u5f8b\uff0c\u5e0c\u671b\u53ef\u4ee5\u4e3a\u57ce\u5e02\u4ea4\u901a\u7684\u53ef\u9760\u6027\u7ba1\u7406\u63d0\u4f9b\u65b0\u9014\u5f84\u3002<\/span><\/section>\n
\"\u96c6\u667a\"<\/span><\/section>\n

\u7b80\u4ecb\uff1a<\/span><\/strong><\/p>\n

\u674e\u5927\u5e86\uff0c\u5317\u4eac\u822a\u7a7a\u822a\u5929\u5927\u5b66\u7814\u7a76\u5458\u3001\u535a\u58eb\u751f\u5bfc\u5e08\u3001\u5317\u822a\u9996\u5c4a\u6821\u52a1\u59d4\u5458\u4f1a\u59d4\u5458\u3002\u56fd\u5bb6\u4f18\u79c0\u9752\u5e74\u79d1\u5b66\u57fa\u91d1\u83b7\u5f97\u8005\u3002\u62c5\u4efb\u4e86\u4e2d\u56fd\u7cfb\u7edf\u5de5\u7a0b\u5b66\u4f1a\u7cfb\u7edf\u53ef\u9760\u6027\u4e13\u59d4\u4f1a\u526f\u79d8\u4e66\u957f\uff0c\u4e2d\u56fd\u4f18\u9009\u6cd5\u7edf\u7b79\u6cd5\u4e0e\u7ecf\u6d4e\u6570\u5b66\u7814\u7a76\u4f1a\u5de5\u4e1a\u5de5\u7a0b\u5206\u4f1a\u5e38\u52a1\u7406\u4e8b\u3002<\/span><\/p>\n

\u8fd1\u5e74\u6765\uff0c\u56f4\u7ed5\u590d\u6742\u7cfb\u7edf\u7684\u53ef\u9760\u6027\u7ba1\u7406\uff0c\u4ee5\u7b2c\u4e00\u4f5c\u8005\u6216\u901a\u8baf\u4f5c\u8005\u5728PNAS\u3001Nature Physics\u3001Nature Communications\u3001RESS\u3001Phys. Rev. Lett.\u7b49\u56fd\u9645\u8457\u540d\u671f\u520a\u4e0a\u53d1\u8868\u7814\u7a76\u6210\u679c\uff1b\u4e3b\u6301\u4e86\u5305\u62ec\u56fd\u5bb6\u81ea\u7136\u79d1\u5b66\u57fa\u91d1\uff0c\u9884\u7814\u9879\u76ee\u548c\u9884\u7814\u91cd\u70b9\u57fa\u91d1\u7b49\u9879\u76ee\u3002<\/span><\/p>\n

\n

\n

\u5f20\u6c5f \u5317\u4eac\u5e08\u8303\u5927\u5b66<\/strong><\/p>\n

\n

\u9898\u76ee\uff1a<\/span><\/strong><\/p>\n

\u4ece\u7b80\u5355\u89c4\u5219\u5230\u673a\u5668\u5b66\u4e60\u2014\u2014\u590d\u6742\u7cfb\u7edf\u5efa\u6a21\u65b9\u6cd5\u7684\u8fdb\u5316<\/span><\/p>\n

\u6458\u8981\uff1a<\/span><\/strong><\/p>\n

\u590d\u6742\u7cfb\u7edf\u5927\u591a\u53ef\u4ee5\u88ab\u770b\u4f5c\u4e00\u4e2a\u76f8\u4e92\u4f5c\u7528\u7684\u7f51\u7edc\u3002\u4f20\u7edf\u7684\u5efa\u6a21\u65b9\u6cd5\u901a\u8fc7\u7b80\u5355\u89c4\u5219\u6765\u62bd\u8c61\u4e2a\u4f53\u4e4b\u95f4\u7684\u76f8\u4e92\u4f5c\u7528\uff0c\u4ece\u800c\u590d\u73b0\u51fa\u7cfb\u7edf\u7684\u5b8f\u89c2\u7279\u6027\u3002\u8fd9\u79cd\u65b9\u6cd5\u867d\u7136\u53ef\u4ee5\u5e2e\u52a9\u5efa\u6a21\u8005\u6d1e\u5bdf\u7cfb\u7edf\u80cc\u540e\u7684\u6df1\u5c42\u673a\u5236\uff0c\u4f46\u8f83\u96be\u7ed3\u5408\u7cfb\u7edf\u7684\u5fae\u89c2\u6570\u636e\u548c\u53c2\u6570\u3002\u968f\u7740\u8fd1\u5e74\u673a\u5668\u5b66\u4e60\u6280\u672f\u7684\u53d1\u5c55\uff0c\u6211\u4eec\u5f00\u53d1\u4e86\u4e00\u79cd\u6839\u636e\u7cfb\u7edf\u7684\u65f6\u95f4\u5e8f\u5217\u800c\u81ea\u52a8\u5efa\u6a21\u7684\u7b97\u6cd5\u6846\u67b6\u3002\u8fd9\u79cd\u65b9\u6cd5\u4e0d\u4ec5\u80fd\u62df\u5408\u7cfb\u7edf\u6f14\u5316\u7684\u52a8\u529b\u5b66\uff0c\u800c\u4e14\u53ef\u4ee5\u81ea\u52a8\u5b66\u51fa\u7cfb\u7edf\u80cc\u540e\u7684\u76f8\u4e92\u4f5c\u7528\u7f51\u7edc\u3002\u6700\u540e\uff0c\u6211\u4eec\u5c06\u7b80\u5355\u89c4\u5219\u548c\u673a\u5668\u5b66\u4e60\u8fd9\u4e24\u79cd\u65b9\u6cd5\u8fdb\u884c\u4e86\u5bf9\u6bd4\u3002<\/span><\/p>\n

\n

\"\u96c6\u667a\"<\/span><\/p>\n

\n

\u7b80\u4ecb\uff1a<\/span><\/strong><\/p>\n

\u5f20\u6c5f\uff0c\u5317\u4eac\u5e08\u8303\u5927\u5b66\u7cfb\u7edf\u79d1\u5b66\u5b66\u9662\u6559\u6388\u3001\u96c6\u667a\u4ff1\u4e50\u90e8\u521b\u59cb\u4eba\u3001\u96c6\u667a\u5b66\u56ed\u521b\u59cb\u4eba\u3001\u66fe\u4efb\u817e\u8baf\u7814\u7a76\u9662\u7279\u8058\u987e\u95ee\uff0c\u590d\u6742\u7cfb\u7edf\u3001\u4eba\u5de5\u667a\u80fd\u7684\u7814\u7a76\u8005\u4e0e\u5e03\u9053\u8005\u3002\u4e3b\u8981\u7814\u7a76\u9886\u57df\u5305\u62ec\u590d\u6742\u7f51\u7edc\u4e0e\u673a\u5668\u5b66\u4e60\u3001\u590d\u6742\u7cfb\u7edf\u5206\u6790\u4e0e\u5efa\u6a21\u3001\u8ba1\u7b97\u793e\u4f1a\u79d1\u5b66\u7b49\u3002\u66fe\u5728Nature Communications\u3001Scientific Reports\u3001Physical Review E\u3001Journal of Theoretical Biology\u7b49\u56fd\u9645\u77e5\u540d\u520a\u7269\u4e0a\u53d1\u8868\u8fc7\u5b66\u672f\u8bba\u6587\u6570\u5341\u7bc7\u3002\u51fa\u7248\u4e13\u8457\u300a\u6570\u5b57\u521b\u4e16\u7eaa\u2014\u2014\u4eba\u5de5\u751f\u547d\u7684\u65b0\u79d1\u5b66\u300b\u3001\u8bd1\u8457\u300a\u81ea\u7136\u4e0e\u4eba\u5de5\u7cfb\u7edf\u4e2d\u7684\u9002\u5e94\u300b\u3001\u6821\u8bd1\u300a\u89c4\u6a21\u300b\u7b49\u8457\u4f5c\uff1b\u66fe\u4e3b\u6301\u300a\u4e92\u8054\u7f51\u4e0a\u7684\u96c6\u4f53\u6ce8\u610f\u529b\u6d41\u7814\u7a76\u300b\u3001\u300a\u52a0\u6743\u6709\u5411\u98df\u7269\u7f51\u7684\u5f02\u901f\u6807\u5ea6\u5f8b\u7814\u7a76\u300b\u7b49\u591a\u9879\u56fd\u5bb6\u7ea7\u79d1\u7814\u9879\u76ee\u3002\u66fe\u591a\u6b21\u56e0\u5408\u4f5c\u7814\u7a76\u800c\u51fa\u8bbf\u8fc7\u56fd\u9645\u77e5\u540d\u7814\u7a76\u9662\u6240\u6216\u9ad8\u6821\uff0c\u5305\u62ec\u7f8e\u56fd\u5723\u5854\u83f2\u7814\u7a76\u6240\uff08Santa Fe Institute\uff09\u3001\u4e9a\u5229\u6851\u90a3\u5dde\u7acb\u5927\u5b66\uff08Arizona State University\uff09\u3001\u5bc6\u897f\u6839\u5927\u5b66\uff08Michigan University\uff09\u7b49\u3002\u5176\u5f00\u521b\u7684\u96c6\u667a\u4ff1\u4e50\u90e8\u662f\u56fd\u5185\u5916\u77e5\u540d\u7684\u5b66\u672f\u793e\u533a\uff0c\u81f4\u529b\u4e8e\u590d\u6742\u7cfb\u7edf\u3001\u4eba\u5de5\u667a\u80fd\u7b49\u591a\u9886\u57df\u7684\u8de8\u5b66\u79d1\u4ea4\u6d41\u4e0e\u5408\u4f5c\u3002<\/span><\/p>\n

\n

\u5f20\u6f58<\/strong><\/p>\n

\u4e2d\u79d1\u9662\u7406\u8bba\u7269\u7406\u7814\u7a76\u6240<\/strong><\/p>\n

\n

\u9898\u76ee\uff1a<\/span><\/strong><\/p>\n

Phase transitions and optimal graph convolution networks in the graph semi-supervised classifications<\/span><\/p>\n

\u6458\u8981\uff1a<\/span><\/strong><\/p>\n

We present statistical-physics studies of the semi-supervised learning on graph-structured data. By analyzing generative model for random networks with discrete labels and bipartite features, we perform an asymptotically exact analysis of the semi-supervised learning problems using the cavity method of statistical physics, and unvill phase transitions beyond which the classification is not possible. Our theory naturally translates to a graph convolution neural network algorithm which is optimal in the underlying generative model which outperforms existing graph convolution network algorithms by a significant margin in the synthetic networks. When applied to real-world semi-supervised classification problems such as citation networks, ouralgorithm achieves comparable performance with the state-of-the art graph convolution neural net-works. Our overall approach can also be used in benchmarking black-box neural-network based models and understanding their strengths as well as limitations.<\/span><\/p>\n

\n

\"\u96c6\u667a\"<\/section>\n

\n

\u7b80\u4ecb\uff1a<\/span><\/strong><\/p>\n

\u5f20\u6f58\uff0c\u672c\u79d1(2004\u5e74), \u535a\u58eb(2009\u5e74)\u6bd5\u4e1a\u4e8e\u5170\u5dde\u5927\u5b66\uff0c\u5176\u540e\u5728\u610f\u5927\u5229\uff0c\u6cd5\u56fd\uff0c\u4ee5\u53ca\u7f8e\u56fd\u5723\u5854\u83f2\u7814\u7a76\u6240\u505a\u535a\u58eb\u540e\u7814\u7a76\u30022015\u5e74\u5165\u9009\u4e2d\u56fd\u79d1\u5b66\u9662\u201c\u767e\u4eba\u8ba1\u5212\u201d\uff0c\u540c\u5e74\u56de\u56fd\u5728\u4e2d\u56fd\u79d1\u5b66\u9662\u7406\u8bba\u7269\u7406\u7814\u7a76\u6240\u5de5\u4f5c\uff0c\u4efb\u526f\u7814\u7a76\u5458\u3001\u7814\u7a76\u5458\u3002\u5f20\u6f58\u7684\u7814\u7a76\u65b9\u5411\u4e3a\u7edf\u8ba1\u7269\u7406\u4e0e\u673a\u5668\u5b66\u4e60\u7684\u4ea4\u53c9\u9886\u57df\uff0c\u8fd1\u5e74\u7684\u7814\u7a76\u5174\u8da3\u96c6\u4e2d\u5728\u7edf\u8ba1\u63a8\u65ad\u95ee\u9898\u4e2d\u7684\u7edf\u8ba1\u7269\u7406\u7406\u8bba\uff0c\u4ee5\u53ca\u57fa\u4e8e\u91cf\u5b50\u548c\u7edf\u8ba1\u7269\u7406\u7684\u975e\u76d1\u7763\u673a\u5668\u5b66\u4e60\u65b0\u65b9\u6cd5\u3002<\/span><\/p>\n

\n

\u675c\u6587\u535a \u5317\u4eac\u822a\u7a7a\u822a\u5929\u5927\u5b66<\/strong><\/p>\n

\n

\u9898\u76ee\uff1a<\/strong><\/span><\/p>\n

\u7a7a\u57df\u590d\u6742\u5ea6\u8bc4\u4f30<\/span><\/p>\n

\n

\"\u96c6\u667a\"<\/span><\/section>\n

\u7b80\u4ecb\uff1a<\/span><\/strong><\/p>\n

\u675c\u6587\u535a\uff0c\u7537\uff0c1985\u5e74\u751f\u30022005\u5e74\u30012011\u5e74\u6bd5\u4e1a\u4e8e\u4e2d\u56fd\u79d1\u5b66\u6280\u672f\u5927\u5b66\u8ba1\u7b97\u673a\u79d1\u5b66\u4e0e\u6280\u672f\u5b66\u9662\uff0c\u5206\u522b\u83b7\u5de5\u5b66\u5b66\u58eb\u3001\u535a\u58eb\u5b66\u4f4d\u30022011\u5e74\u52a0\u5165\u5317\u4eac\u822a\u7a7a\u822a\u5929\u5927\u5b66\u7535\u5b50\u4fe1\u606f\u5de5\u7a0b\u5b66\u9662\u5e76\u76f4\u8bc4\u526f\u6559\u6388\u3002\u7814\u7a76\u65b9\u5411\u5305\u62ec\u7f51\u7edc\u79d1\u5b66\u3001\u667a\u80fd\u8ba1\u7b97\u3001\u6001\u52bf\u8ba1\u7b97\u3001\u822a\u7a7a\u4ea4\u901a\u8fd0\u884c\u8c03\u63a7\u7b49\u3002\u4e3b\u6301\u56fd\u5bb6\u81ea\u7136\u79d1\u5b66\u57fa\u91d12\u9879\u3001\u56fd\u5bb6\u79d1\u6280\u652f\u6491\u8ba1\u52121\u9879\uff0c\u53c2\u4e0e863\u8bfe\u9898\u3001973\u8bfe\u9898\u3001\u56fd\u5bb6\u79d1\u6280\u652f\u6491\u8ba1\u5212\u3001\u81ea\u7136\u79d1\u5b66\u57fa\u91d1\u521b\u65b0\u7814\u7a76\u7fa4\u4f53\u7b49\u91cd\u5927\u91cd\u70b9\u9879\u76ee\uff1b\u5728Applied Mathematics and Computation\u3001Scientific Reports\u3001IEEE TCAS-ii\u3001EPL\u7b49\u56fd\u9645\u5b66\u672f\u520a\u7269\u4e0a\u53d1\u8868SCI\u8bba\u658740\u4f59\u7bc7\uff0c\u7533\u8bf7\u56fd\u5bb6\u53d1\u660e\u4e13\u522910\u4f59\u9879\u3002\u83b72015\u5e74\u5ea6\u4e2d\u56fd\u7535\u5b50\u5b66\u4f1a\u81ea\u7136\u79d1\u5b66\u4e00\u7b49\u5956\uff08\u6392\u540d\u7b2c2\uff09\u30012016\u5e74\u5ea6\u4e2d\u56fd\u667a\u80fd\u4ea4\u901a\u5b66\u4f1a\u79d1\u5b66\u6280\u672f\u4e00\u7b49\u5956\uff08\u6392\u540d\u7b2c2\uff09\uff0c2015\u5e74\u5165\u9009\u4e2d\u56fd\u79d1\u534f\u9996\u5c4a\u201c\u9752\u5e74\u4eba\u624d\u6258\u4e3e\u5de5\u7a0b\u201d\u3002<\/span><\/p>\n<\/div>\n

\u6458\u8981\uff1a<\/p>\n

\u7a7a\u57df\u590d\u6742\u5ea6\u662f\u8861\u91cf\u7a7a\u57df\u5b89\u5168\u8fd0\u884c\u7684\u91cd\u8981\u6307\u6807\uff0c\u4e3a\u4fdd\u8bc1\u6574\u4e2a\u7a7a\u7ba1\u7cfb\u7edf\u9ad8\u6548\u8fd0\u884c\uff0c\u73b0\u6709\u7684\u624b\u6bb5\u662f\u5c06\u7a7a\u57df\u5212\u5206\u4e3a\u591a\u4e2a\u57fa\u672c\u7684\u7ba1\u5236\u5355\u5143\u2014\u2014\u6247\u533a\uff0c\u91c7\u53d6\u52a8\u6001\u5212\u5206\u3001\u4ea4\u901a\u6d41\u91cf\u7ba1\u7406\u7b49\u8fd0\u884c\u8c03\u63a7\u63aa\u65bd\u6765\u4fdd\u8bc1\u5404\u4e2a\u6247\u533a\u7684\u4ea4\u901a\u590d\u6742\u7a0b\u5ea6\u5728\u7ba1\u5236\u5458\u6240\u80fd\u628a\u63a7\u7684\u8303\u56f4\u5185\u3002\u7136\u800c\uff0c\u5f71\u54cd\u7a7a\u57df\u590d\u6742\u5ea6\u7684\u9759\/\u52a8\u6001\u5143\u7d20\u4f17\u591a\u4e14\u8026\u5408\u590d\u6742\uff0c\u96be\u4ee5\u8fdb\u884c\u51c6\u786e\u8bc4\u4f30\u3002\u8fd1\u5e74\u6765\uff0c\u4e00\u4e9b\u5b66\u8005\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\uff0c\u57fa\u4e8e\u5927\u6837\u672c\u6316\u6398\u6765\u6784\u5efa\u6570\u5341\u79cd\u7a7a\u4e2d\u4ea4\u901a\u590d\u6742\u6027\u56e0\u5b50\u4e0e\u590d\u6742\u5ea6\u4e4b\u95f4\u7684\u5173\u8054\uff0c\u53ef\u8fbe\u5230\u8f83\u9ad8\u7684\u7a7a\u57df\u590d\u6742\u5ea6\u8ba1\u7b97\u51c6\u786e\u5ea6\u3002\u4f46\u7531\u4e8e\u5404\u6761\u6837\u672c\u5bf9\u5e94\u7684\u7a7a\u57df\u590d\u6742\u5ea6\u7b49\u7ea7\u9700\u8981\u7ba1\u5236\u4e13\u5bb6\u53c2\u4e0e\u6807\u5b9a\uff0c\u9020\u6210\u6837\u672c\u6536\u96c6\u6240\u9700\u7684\u4eba\u529b\u53ca\u65f6\u95f4\u6210\u672c\u8f83\u9ad8\uff0c\u53e6\u5916\uff0c\u7a7a\u57df\u7ed3\u6784\u7684\u52a8\u6001\u6f14\u53d8\u4f1a\u5bfc\u81f4\u5df2\u6807\u5b9a\u6837\u672c\u8fc7\u65f6\u3002\u672c\u62a5\u544a\u4ece\u5c0f\u6837\u672c\u73af\u5883\u4e0b\u7684\u7a7a\u57df\u590d\u6742\u5ea6\u8bc4\u4f30\u95ee\u9898\u51fa\u53d1\uff0c\u57fa\u4e8e\u4e00\u79cd\u77e5\u8bc6\u8fc1\u79fb\u7684\u5b66\u4e60\u6846\u67b6\u8ba1\u7b97\u7a7a\u57df\u590d\u6742\u5ea6\uff0c\u4e3a\u8fdb\u4e00\u6b65\u964d\u4f4e\u4e13\u5bb6\u6807\u7b7e\u7684\u5f71\u54cd\uff0c\u8bbe\u8ba1\u4e86\u4e00\u79cd\u57fa\u4e8e\u6df1\u5ea6\u65e0\u76d1\u7763\u5b66\u4e60\u7684\u7a7a\u57df\u590d\u6742\u5ea6\u8ba1\u7b97\u6a21\u578b\uff0c\u80fd\u591f\u5728\u6ca1\u6709\u4e13\u5bb6\u6807\u7b7e\u7684\u6761\u4ef6\u4e0b\u5145\u5206\u6316\u6398\u6837\u672c\u6570\u636e\u4e2d\u8574\u542b\u7684\u7a7a\u57df\u590d\u6742\u5ea6\u4fe1\u606f\uff0c\u8fdb\u800c\u5bf9\u7a7a\u57df\u6247\u533a\u6570\u636e\u8fdb\u884c\u51c6\u786e\u7684\u590d\u6742\u5ea6\u5206\u7c7b\u3002<\/p>\n

\n

\n

\n

\u97e9\u7b71\u749e \u676d\u5dde\u5e08\u8303\u5927\u5b66<\/strong><\/p>\n

\n

\u9898\u76ee\uff1a<\/span><\/strong><\/p>\n

\u8054\u5408\u7ecf\u6d4e\uff1a\u4fe1\u606f\u65f6\u4ee3\u7ecf\u6d4e\u53d1\u5c55\u7684\u5168\u65b0\u8303\u5f0f<\/span><\/p>\n

\u6458\u8981\uff1a<\/span><\/strong><\/p>\n

\u57fa\u4e8e\u7ecf\u6d4e\u590d\u6742\u6027\u89d2\u5ea6\uff0c\u5bf9\u5f53\u524d\u4ee5\u4e92\u8054\u7f51\u65b0\u7ecf\u6d4e\u6a21\u5f0f\u4e3a\u4ee3\u8868\u7684\u5404\u7c7b\u7ecf\u6d4e\u53d1\u5c55\u8d8b\u52bf\u8fdb\u884c\u68b3\u7406\u5206\u6790\uff0c\u63d0\u51fa\u4e86\u4e00\u7c7b\u7531\u793e\u4f1a\u6210\u5458\u95f4\u7684\u81ea\u7531\u8054\u5408\u6765\u9a71\u52a8\u793e\u4f1a\u751f\u4ea7\u7684\u8054\u5408\u7ecf\u6d4e\u5f62\u6001\uff0c\u5b83\u4ee3\u8868\u7740\u4fe1\u606f\u65f6\u4ee3\u7ecf\u6d4e\u53d1\u5c55\u7684\u4e00\u79cd\u5168\u65b0\u8303\u5f0f\u3002\u8fdb\u4e00\u6b65\u5730\uff0c\u672c\u6587\u5bf9\u8054\u5408\u7ecf\u6d4e\u7684\u57fa\u672c\u5f62\u6001\u8fdb\u884c\u4e86\u5168\u65b9\u4f4d\u7684\u9884\u6d4b\u6027\u63cf\u8ff0\uff0c\u9610\u660e\u8054\u5408\u7ecf\u6d4e\u662f\u4e00\u79cd\u5168\u65b0\u7684\u7ecf\u6d4e\u5f62\u6001\uff0c\u5b83\u4ee5\u81ea\u7531\u8054\u5408\u7684\u5f00\u653e\u793e\u533a\u4e3a\u4e3b\u8981\u751f\u4ea7\u7ec4\u7ec7\u5f62\u5f0f\uff0c\u690d\u6839\u4e8e\u673a\u5668\u751f\u6001\uff0c\u76f4\u9762\u53d1\u5c55\u6027\u8bc9\u6c42\uff0c\u9002\u5e94\u4e8e\u9ad8\u5ea6\u590d\u6742\u7684\u7ecf\u6d4e\u73af\u5883\uff0c\u5176\u793e\u4f1a\u8d44\u6e90\u5168\u9762\u5f00\u653e\uff0c\u793e\u4f1a\u516c\u5e73\u7a0b\u5ea6\u8f83\u9ad8\uff0c\u53ef\u4ee5\u89e3\u51b3\u5f53\u524d\u7ecf\u6d4e\u73af\u5883\u4e0b\u7684\u8bb8\u591a\u6df1\u5c42\u77db\u76fe\u6027\u95ee\u9898\u3002\u8fd9\u79cd\u7ecf\u6d4e\u5f62\u6001\u7684\u51fa\u73b0\u548c\u5efa\u7acb\uff0c\u6807\u5fd7\u7740\u81ea\u5546\u54c1\u7ecf\u6d4e\u7a81\u7834\u81ea\u7136\u7ecf\u6d4e\u4ee5\u6765\u7684\u53c8\u4e00\u6b21\u4eba\u7c7b\u53d1\u5c55\u8303\u5f0f\u7684\u5927\u8f6c\u79fb\u3002<\/span><\/p>\n

\n

\"\u96c6\u667a\"<\/section>\n

 <\/p>\n

\u7b80\u4ecb\uff1a<\/span><\/strong><\/p>\n

\u97e9\u7b71\u749e\uff0c\u676d\u5dde\u5e08\u8303\u5927\u5b66\u4fe1\u606f\u7ecf\u6d4e\u7814\u7a76\u6240\u526f\u6559\u6388\u30022012\u5e74\u4e8e\u4e2d\u56fd\u79d1\u5b66\u6280\u672f\u5927\u5b66\u83b7\u5f97\u535a\u58eb\u5b66\u4f4d\uff0c\u4e3b\u653b\u590d\u6742\u6027\u79d1\u5b66\u4e0e\u793e\u4f1a\u7269\u7406\u4ea4\u53c9\u5b66\u79d1\u7814\u7a76\uff0c\u5176\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u5305\u542b\u4e86\u4eba\u7c7b\u884c\u4e3a\u7684\u7edf\u8ba1\u7279\u6027\u7684\u5b9e\u8bc1\u4e0e\u52a8\u529b\u5b66\u673a\u5236\u3001\u793e\u4f1a\u7cfb\u7edf\u4e2d\u7684\u590d\u6742\u6027\u73b0\u8c61\u53ca\u5176\u52a8\u529b\u5b66\u673a\u5236\u7b49\u3002\u5148\u540e\u5728\u56fd\u5185\u5916\u5b66\u672f\u671f\u520a\u4e0a\u53d1\u8868\u7814\u7a76\u8bba\u658760\u4f59\u7bc7\uff0c\u5176\u4e2dSCI\u6536\u5f5540\u4f59\u7bc7\uff0cSSCI\u6536\u5f558\u7bc7\uff0c\u53c2\u52a0\u4e13\u8457\u7f16\u5199\u4e24\u90e8\u548c\u4ea7\u4e1a\u62a5\u544a\u4e00\u90e8\uff0c\u5148\u540e\u4e3b\u6301\u56fd\u5bb6\u81ea\u7136\u79d1\u5b66\u57fa\u91d1\u9879\u76ee\u4e24\u9879\u3002<\/span><\/p>\n

\n

\u674e\u777f\u742a \u5317\u4eac\u5316\u5de5\u5927\u5b66<\/strong><\/p>\n

\n

\u9898\u76ee\uff1a<\/span><\/strong><\/p>\n

Urban Spatial Scaling: From active population to folding cities<\/span><\/p>\n

\u6458\u8981\uff1a<\/span><\/strong><\/p>\n

With data from various sources, we are able to better sense the activities of residents and the pulse of the whole city. Based on cell phone data and proper algorithms, we can get accurate trajectories of individuals, reveal the connection between house price and air pollution exposure of residents, and get active population (AP) distribution as well as the OD matrix of the city. By considering the travel demand together with features of the road network, we propose an indicator named population-weighted efficiency (PWE) which can give a more comprehensive and systematic evaluation of the efficiency of the transportation network, and detect 3 typical types of inefficient routes. In China, surrounding working units, residential area, universities, governments with walls (i.e., compounds or \u201cbig yard\u201d) are quite common and unique, which is designed to an independent ecosystem, but nowadays can cause inefficiency of the city, and inconvenience to residents, and may have some other profound effects. And from a dynamical and interactive perspective, based on the concept of AP, we come up with a simple model (1 free parameter and 4 simple assumptions) to reproduce the spatial distribution within cities and explain the origins of spatial scaling within cities and scaling laws across cities at the same time. With the development of building and transportation technologies, the city sprawl itself in both 2D (fast routes) and 3D (skyscrapers) space, which is the old story we have been hearing for decades. From a brand new perspective, such processes are actually folding more space to the urban central area to further promotes human interactions. From such a folding cities perspective, we can better reveal the true structure of urban systems and change how we perceive our living environment.<\/span><\/p>\n

\n

\"\u96c6\u667a\"<\/section>\n

\u7b80\u4ecb\uff1a<\/span><\/strong><\/p>\n

\n

\u674e\u777f\u742a\uff0c\u5317\u4eac\u5316\u5de5\u5927\u5b66\u4fe1\u606f\u79d1\u5b66\u4e0e\u6280\u672f\u5b66\u9662\u526f\u6559\u6388\uff0c\u57ce\u5e02\u7f51\u7edc\u5b9e\u9a8c\u5ba4\u4e3b\u4efb\uff0c\u5317\u4eac\u5e08\u8303\u5927\u5b66\u7cfb\u7edf\u79d1\u5b66\u535a\u58eb\uff0c\u535a\u58eb\u671f\u95f4\u5728\u9ebb\u7701\u7406\u5de5\u5b66\u9662\u4e0e\u6ce2\u58eb\u987f\u5927\u5b66\u8fdb\u884c\u8054\u5408\u57f9\u517b\uff0c\u672c\u79d1\u5c31\u8bfb\u4e8e\u7535\u5b50\u79d1\u6280\u5927\u5b66\u56fd\u9645\u5316\u8f6f\u4ef6\u4eba\u624d\u5b9e\u9a8c\u73ed\u3002\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u4e3a\u57ce\u5e02\u5927\u6570\u636e\u5206\u6790\u4e0e\u5efa\u6a21\u3001\u793e\u4f1a\u7f51\u7edc\u7814\u7a76\u4e0e\u6d41\u884c\u75c5\u4f20\u64ad\u52a8\u529b\u5b66\u3002\u76ee\u524d\u53d1\u8868SCI\u8bba\u6587\u5341\u4f59\u7bc7\uff0c\u5f15\u7528150\u4f59\u6b21\uff0c\u76f8\u5e94\u5de5\u4f5c\u53d1\u8868\u5728Nature Communications\u3001Scientific Reports\u3001PLoS One\u3001Physica A\u7b49SCI\u671f\u520a\uff0c\u5e76\u5728Statphys25\/26\u3001Conference of Complex Systems\u3001NetSci\u3001NetSciX\u7b49\u591a\u4e2a\u56fd\u9645\u5927\u4f1a\u4e0a\u4f5c\u53e3\u5934\u62a5\u544a\uff0c\u66fe\u8363\u83b7\u7b2c\u5341\u4e8c\u5c4a\u793e\u4f1a\u7f51\u4e0e\u793e\u4f1a\u8d44\u672c\u7814\u7a76\u5e74\u4f1a\u6700\u4f73\u8bba\u6587\u5956\u3002\u76ee\u524d\u662fScientific Reports\u3001Habitat International\u3001Cities\u3001IEEE Access\u3001PLoS One\u3001Physica A\u3001IJMPC\u3001TRB\/TRR\u3001Journal of Systems Science and Complexity\u7b49\u591a\u4e2aSCI\u4e0eSSCI\u671f\u520a\u7684\u5ba1\u7a3f\u4eba\u3002<\/span><\/p>\n

\n

\u9152\u5e97\u4fe1\u606f\uff1a<\/strong><\/p>\n

 <\/p>\n

1. Yitel <\/strong>\u548c\u9890\u9152\u5e97(\u8f6f\u4ef6\u56ed5\u53f7\u8def)*The Nearest Hotel (1.5 km)<\/strong><\/span><\/section>\n
Address:ZPark Building 9, No. 8 West Dongbeiwang Road, Haidian District, Beijing, 100094, China <\/span><\/section>\n
\u5317\u4eac\u5e02\u6d77\u6dc0\u533a\u4e1c\u5317\u65fa\u897f\u8def8\u53f7\u4e2d\u5173\u6751\u8f6f\u4ef6\u56ed9\u53f7\u697c (\u4e2d\u5173\u6751\u8f6f\u4ef6\u56ed\u5357\u95e8)<\/span><\/section>\n
Telephone: +86-10-82826677 \u6216\u5b8b\u7ecf\u7406\uff1a<\/span>15503059199<\/span><\/section>\n
\u534f\u8bae\u4ef7\uff1a<\/span>\u6807\u95f4\/\u5927\u5e8a\uff1a<\/span>480\u5143\/\u665a<\/span><\/section>\n
2.<\/strong>Xinxing Hotel \u946b\u661f\u9152\u5e97 *Domestic guests only <\/strong><\/span><\/section>\n
Address: \u5317\u4eac\u5e02\u6d77\u6dc0\u533a\u4e1c\u5317\u65fa\u8def\u516b\u7ef4\u5b66\u68212\u53f7\u697c<\/span><\/section>\n
Telephone: +86-10-83399988<\/span><\/section>\n
\u534f\u8bae\u4ef7\uff1a<\/span>\u6807\u95f4\/\u5927\u5e8a\uff1a<\/span>388\u5143\/\u665a<\/span><\/section>\n
3. H<\/strong>oliday Inn Express Shangdi Beijing \u5317\u4eac\u4e0a\u5730\u667a\u9009\u5047\u65e5\u9152\u5e97<\/strong><\/span><\/section>\n
Address: No. 33 Shangdi East Road, Haidian District, Beijing, 100085, China<\/span><\/section>\n
\u5317\u4eac\u5e02\u6d77\u6dc0\u533a\u4e0a\u5730\u4e1c\u8def33\u53f7 <\/span><\/section>\n
Telephone: +86-10-82709999 \/ 400-830-2360\u6216Vivian\uff1a<\/span>18611298116<\/span><\/section>\n
\u534f\u8bae\u4ef7\uff1a<\/span>\u6807\u95f4\/\u5927\u5e8a\uff1a<\/span>528\u5143\/\u665a<\/span><\/section>\n
4. <\/strong>Tianxinliang Hotel \u5929\u4fe1\u4eae\u9152\u5e97<\/strong><\/span><\/section>\n
Address:29B, No.11 Anningzhuang Road, Haidian District, Beijing<\/span><\/section>\n
\u5317\u4eac\u5e02\u6d77\u6dc0\u533a\u5b89\u5b81\u5e84\u8def11\u53f729B<\/span><\/section>\n
Telephone:\u59dc\u7ecf\u7406\uff1a<\/span>18613808343<\/span><\/section>\n
\u534f\u8bae\u4ef7\uff1a<\/span>\u6807\u95f4442\/\u665a\uff0c\u5927\u5e8a468\/\u665a<\/span><\/section>\n
5. <\/strong>Hanting Hotel \u6c49\u5ead\u9152\u5e97(\u5317\u4eac\u4e0a\u5730\u73af\u5c9b\u5e97) *Domestic guests only<\/strong><\/span><\/section>\n
Address: \u5317\u4eac\u5e02\u6d77\u6dc0\u533a\u4e0a\u5730\u516d\u885717\u53f7(\u9760\u8fd1\u4e0a\u5730\u897f\u8def\uff0c\u5eb7\u5f97\u5927\u53a6\u897f\u697c)<\/span><\/section>\n
Telephone: +86-10-82897070<\/span><\/section>\n
6. Ho<\/strong>me Inn \u5982\u5bb6\u9152\u5e97(\u5317\u4eac\u4e0a\u5730\u5e97)<\/strong><\/span><\/section>\n
Address: No. 11 Shangdi Chuangye Road, Haidian District, Beijing<\/span><\/section>\n
\u5317\u4eac\u5e02\u6d77\u6dc0\u533a\u4e0a\u5730\u521b\u4e1a\u8def11\u53f7<\/span><\/section>\n
Telephone: +86-010-62962299 \/ 400-820-3333<\/span><\/section>\n
7. 7 D<\/strong>ays Inn7\u5929\u8fde\u9501\u9152\u5e97 *Domestic guests only<\/strong><\/span><\/section>\n
Address: \u5317\u4eac\u5e02\u6d77\u6dc0\u533a\u4e0a\u5730\u516d\u885728\u53f7 (\u4e0a\u5730\u516d\u8857\u4e0e\u4e0a\u5730\u897f\u8def\u4ea4\u53c9\u53e3)<\/span><\/section>\n
Telephone: +86-10-82783388<\/span><\/section>\n
\u8bf4\u660e\uff1a<\/span>\u524d\u56db\u4e2a\u662f\u4f1a\u8bae\u7684\u534f\u8bae\u9152\u5e97\uff0c\u5176\u4e2d\u548c\u9890\u3001\u4e0a\u5730\u667a\u9009\u548c\u5929\u4fe1\u4eae\u4e09\u4e2a\u9152\u5e97\u8bf7\u77ed\u4fe1\u8054\u7cfb\u9500\u552e\u7ecf\u7406\u8fdb\u884c\u9884\u8ba2<\/span><\/strong><\/section>\n
\"\u96c6\u667a\"<\/section>\n

\n

\"\u96c6\u667a\"<\/section>\n
\u626b\u7801\u586b\u5199\u62a5\u540d\u4fe1\u606f<\/span><\/strong><\/section>\n

\n

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

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

\n

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

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

 <\/p>\n

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

\"\u96c6\u667a\"<\/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

\u539f\u6587\u59cb\u53d1\u4e8e\u5fae\u4fe1\u516c\u4f17\u53f7\uff08\u96c6\u667a\u4ff1\u4e50\u90e8\uff09\uff1a\u96c6\u667a<\/a><\/p><\/blockquote>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

\u5bfc\u8bed \u5317\u4eac\u8ba1\u7b97\u79d1\u5b66\u7814\u7a76\u4e2d\u5fc3\u5c06\u4e8e10\u670827\u65e5\uff08\u5468\u65e5\uff09\u5728\u5317\u4eac\u4e3e\u529e\u5168\u5929\u7684\u590d\u6742\u7cfb\u7edf\u4e0e\u793e\u4f1a\u8ba1\u7b97\u7814\u8ba8\u4f1a\u6d3b\u52a8\uff0c\u672c\u6b21\u6d3b\u52a8\u7531\u4e2d\u6587\u4fe1\u606f\u5b66\u4f1a\u793e\u4f1a\u5a92\u4f53\u5904\u7406\u4e13\u59d4\u4f1a\u793e\u4f1a\u7f51\u7edc\u5206\u6790\u4e13\u4e1a\u7ec4\uff08\u7b79\uff09\u8054\u5408\u7ec4\u7ec7\u3002\u4f1a\u8bae\u8bae\u7a0b\u4f9b\u5404\u4f4d\u540c\u597d\u53c2\u8003\u3002\u6b22\u8fce\u611f\u5174\u8da3\u7684\u670b\u53cb\u62a5\u540d\u53c2\u4e0e\u3002\u96c6\u667a\u4ff1\u4e50\u90e8\u5c06\u8fdb\u884c\u5168\u7a0b\u76f4\u64ad\u3002 \u80cc\u666f\u4ecb\u7ecd \u4f5c\u4e3a\u4e00\u4e2a\u8de8\u5b66\u79d1\u9886\u57df\uff0c\u590d\u6742\u7cfb\u7edf\u79d1\u5b66\u5438\u6536\u4e86\u8bb8\u591a…<\/p>\n","protected":false},"author":1,"featured_media":17288,"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\/17285"}],"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=17285"}],"version-history":[{"count":0,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/17285\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/media\/17288"}],"wp:attachment":[{"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=17285"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=17285"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=17285"},{"taxonomy":"special","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fspecial&post=17285"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}