{"id":3156,"date":"2025-05-15T09:42:07","date_gmt":"2025-05-15T01:42:07","guid":{"rendered":"https:\/\/www.15zhi.net\/blog\/?p=3156"},"modified":"2025-05-15T09:42:07","modified_gmt":"2025-05-15T01:42:07","slug":"202504-%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-transformer-autoencoder-for-k-means-efficient-clustering","status":"publish","type":"post","link":"https:\/\/www.15zhi.net\/blog\/202504-%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-transformer-autoencoder-for-k-means-efficient-clustering\/","title":{"rendered":"202504 \u8bba\u6587\u7814\u8bfb-Transformer Autoencoder for K-means Efficient clustering"},"content":{"rendered":"\n

\u6765\u6e90:Engineering Applications of Artificial Intelligence, 2024<\/p>\n\n\n\n

\u4f5c\u8005: Wenhao Wu, Weiwei Wang, Xixi Jia, Xiangchu Feng<\/p>\n\n\n\n

\u5355\u4f4d:\u897f\u7535\u5927\u5b66\u6570\u5b66\u4e0e\u7edf\u8ba1\u5b66\u9662<\/p>\n\n\n\n

\u4e00\u3001\u8bba\u6587\u4e3b\u8981\u5de5\u4f5c\u53ca\u8d21\u732e<\/h2>\n\n\n\n

\u4e3a\u4e86\u4f18\u5316\u7279\u5f81\u5b66\u4e60\u548cK-Means\u805a\u7c7b\uff0c\u8fd9\u7bc7\u8bba\u6587\u63d0\u51fa\u4e86\u4e00\u4e2a\u65b0\u7684\u6df1\u5c42\u805a\u7c7b\u7f51\u7edc\uff0c\u79f0\u4e3aTAKE\uff0c\u7528\u4e8eK-Means\u7684\u6709\u6548\u805a\u7c7b\u3002<\/p>\n\n\n\n

\u00a0 \u5b83\u7531\u4e24\u4e2a\u6a21\u5757\u7ec4\u6210\uff1a\u7528\u4e8e\u7279\u5f81\u5b66\u4e60\u7684Transformer\u81ea\u52a8\u7f16\u7801\u5668\uff08TAE\uff09\u548c\u7528\u4e8e\u805a\u7c7b\u7684KNet\u3002 TAE\u7ed3\u5408\u4e86Transformer\u7ed3\u6784\uff0c\u4ee5\u5b66\u4e60\u5168\u5c40\u7279\u5f81\u548c\u5bf9\u6bd4\u5ea6\u5b66\u4e60\u673a\u5236\u4ee5\u589e\u5f3a\u7279\u5f81\u6b67\u89c6\u3002KNet\u662f\u901a\u8fc7\u5bf9K-Means\u6a21\u578b\u7684\u52a0\u901f\u6295\u5f71\u68af\u5ea6\u4e0b\u964d\u8fed\u4ee3\u6765\u6784\u5efa\u7684\u3002<\/p>\n\n\n\n

\u00a0 \u8be5\u7f51\u7edc\u5206\u4e3a\u4e24\u4e2a\u9636\u6bb5\uff1a\u9884\u5904\u7406\u548c\u805a\u7c7b\u3002 \u5728\u8bad\u7ec3\u8bad\u7ec3\u4e2d\uff0c\u901a\u8fc7\u6700\u5927\u7a0b\u5ea6\u5730\u51cf\u5c11\u57fa\u4e8e\u4f59\u5f26\u76f8\u4f3c\u6027\u7684\u91cd\u5efa\u635f\u5931\uff0c\u5bf9\u6bd4\u5ea6\u635f\u5931\uff08CL\uff09\u548c\u51f8\u7ec4\u5408\u635f\u5931\uff08CCL\uff09\u6765\u4f18\u5316TAE\u3002 CCL\u9f13\u52b1\u589e\u5f3a\u90bb\u5c45\u6570\u636e\u7684\u529f\u80fd\u3002 \u5728\u805a\u7c7b\u9636\u6bb5\uff0c\u901a\u8fc7\u6700\u5927\u7a0b\u5ea6\u5730\u51cf\u5c11\u91cd\u5efa\u635f\u5931\u548cK-Means\u805a\u7c7b\u635f\u5931\u3002<\/p>\n\n\n\n

\u4e8c\u3001\u6a21\u578b\u6846\u67b6<\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n

2.1. Pretraining loss of TAE<\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n

2.2. Training strategy<\/h2>\n\n\n\n

\u6574\u4e2a\u7f51\u7edc\u91c7\u7528\u5206\u4e3a\u4e24\u4e2a\u9636\u6bb5\uff1a\u9884\u5904\u7406\u548c\u805a\u7c7b\u3002 \u8bad\u7ec3\u9636\u6bb5\u7684\u4e3b\u8981\u76ee\u7684\u662f\u901a\u8fc7\u501f\u52a9\u5bf9\u6bd4\u5ea6\u635f\u5931\u8bad\u7ec3TAE\u6765\u826f\u597d\u7684\u521d\u59cb\u5316\uff0c\u4ee5\u5b66\u4e60\u5224\u522b\u7279\u5f81\u3002 \u805a\u7c7b\u9636\u6bb5\u7aef\u5230\u7aef\u8bad\u7ec3\u6574\u4e2a\u91c7\u96c6\uff0c\u4ee5\u83b7\u5f97\u8f93\u5165\u6570\u636e\u7684\u7fa4\u96c6\u5206\u914d\u3002<\/p>\n\n\n\n

\u4e09\u3001\u5b9e\u9a8c\u7ed3\u679c<\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n

\u56db\u3001\u603b\u7ed3\u4e0e\u601d\u8003<\/h2>\n\n\n\n

\uff08\u4e00\uff09\u3001\u8bba\u6587\u7684\u6838\u5fc3\u5185\u5bb9:<\/p>\n\n\n\n

\u00a0\u00a0\u00a0 \u8fd9\u7bc7\u6587\u7ae0\u63d0\u51fa\u4e86\u4e00\u4e2a\u7aef\u5230\u7aef\u7684\u6df1\u7fa4\u96c6\u7f51\u7edc\u3002 \u5b83\u7531\u4e24\u4e2a\u6a21\u5757\u7ec4\u6210\uff1a\u53d8\u538b\u5668\u81ea\u52a8\u7f16\u7801\u5668\uff08TAE\uff09\u548cKNET\u3002 \u4e3a\u4e86\u63d0\u53d6\u9002\u5408K-\u5747\u503c\u805a\u7c7b\u7684\u533a\u5206\u7279\u5f81\uff0c\u901a\u8fc7\u4f7f\u7528\u5efa\u8bae\u7684\u91cd\u5efa\u635f\u5931\uff0c\u51f8\u7ec4\u5408\u635f\u5931\u548c\u5bf9\u6bd4\u5ea6\u635f\u5931\uff0c\u53ef\u4ee5\u9884\u5148\u8bad\u7ec3TAE\u3002<\/p>\n\n\n\n

\uff08\u4e8c\uff09\u3001\u7efc\u5408\u5bf9\u9f50\u601d\u8003:<\/p>\n\n\n\n

1.\u8fd9\u7bc7\u8bba\u6587\u5728\u805a\u7c7b\u65b9\u9762\u8fd8\u662f\u6839\u690d\u4e8eK-Means\uff0c\u65e0\u6cd5\u6446\u8131\u9884\u5b9a\u4e49\u7c07\u6570\u7684\u5c40\u9650\u6027\u3002<\/p>\n\n\n\n

2.\u6587\u7ae0\u53ea\u662f\u5bf9transformer\u7ed3\u6784\u7684\u7b80\u5355\u5e94\u7528\uff0c\u5e76\u4e14\u591a\u9488\u5bf9\u4e8e\u56fe\u50cf\u6570\u636e\u3002<\/p>\n\n\n\n

<\/p>\n","protected":false},"excerpt":{"rendered":"

\u6765\u6e90:Engineering Applications of Artificial Intelligence, […]<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-3156","post","type-post","status-publish","format-standard","hentry","category-events"],"_links":{"self":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/3156","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/comments?post=3156"}],"version-history":[{"count":2,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/3156\/revisions"}],"predecessor-version":[{"id":3168,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/3156\/revisions\/3168"}],"wp:attachment":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/media?parent=3156"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/categories?post=3156"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/tags?post=3156"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}