{"id":4492,"date":"2024-08-29T16:26:59","date_gmt":"2024-08-29T08:26:59","guid":{"rendered":"https:\/\/nullthought.net\/?p=4492"},"modified":"2025-03-20T14:04:37","modified_gmt":"2025-03-20T06:04:37","slug":"%e5%af%b9%e6%af%94%e5%ad%a6%e4%b9%a0%ef%bc%88contrastive-learning%ef%bc%89%e5%81%a5%e5%ba%b7%e7%9b%b8%e5%85%b3%e5%a3%b0%e5%ad%a6%e4%bf%a1%e5%8f%b7%e7%9a%84%e9%9f%b3%e9%a2%91%e5%a2%9e%e5%bc%ba%e4%bc%98","status":"publish","type":"post","link":"https:\/\/nullthought.net\/?p=4492","title":{"rendered":"\u5bf9\u6bd4\u5b66\u4e60\uff08Contrastive Learning\uff09\u5065\u5eb7\u76f8\u5173\u58f0\u5b66\u4fe1\u53f7\u7684\u97f3\u9891\u589e\u5f3a\u4f18\u5316"},"content":{"rendered":"\n

\u8bba\u6587Optimizing Audio Augmentations for Contrastive Learning of Health-Related Acoustic Signals<\/a><\/strong>\uff08\u300a\u5bf9\u6bd4\u5b66\u4e60\u5065\u5eb7\u76f8\u5173\u58f0\u5b66\u4fe1\u53f7\u7684\u97f3\u9891\u589e\u5f3a\u4f18\u5316\u300b\uff09\uff0c\u7814\u7a76\u7684\u4e3b\u8981\u76ee\u6807\u662f\u6539\u8fdb\u5065\u5eb7\u76f8\u5173\u58f0\u5b66\u4fe1\u53f7\uff08\u4f8b\u5982\u54b3\u55fd\u548c\u547c\u5438\u58f0\uff09\u5728\u5bf9\u6bd4\u5b66\u4e60\u4e2d\u7684\u8868\u73b0\uff0c\u5177\u4f53\u662f\u901a\u8fc7\u4f18\u5316\u97f3\u9891\u589e\u5f3a\u7b56\u7565\u6765\u5b9e\u73b0\u7684\u3002<\/p>\n\n\n\n

\u4f5c\u8005\u4e3aLouis Blankemeier, Sebastien Baur, Wei-Hung Weng, Jake Garrison, Yossi Matias, Shruthi Prabhakara, Diego Ardila, Zaid Nabulsi\uff0c\u5747\u6765\u81eaGoogle Research\u3002<\/p>\n\n\n\n

\u8bba\u6587\u6982\u8981\u5185\u5bb9\u5982\u4e0b\uff1a<\/p>\n\n\n\n

\u4e00\u3001\u80cc\u666f\u4e0e\u52a8\u673a<\/h5>\n\n\n\n

\u5065\u5eb7\u76f8\u5173\u7684\u58f0\u5b66\u4fe1\u53f7\u5982\u54b3\u55fd\u58f0\u548c\u547c\u5438\u58f0\u5bf9\u4e8e\u533b\u5b66\u8bca\u65ad\u548c\u5065\u5eb7\u76d1\u6d4b\u6709\u7740\u91cd\u8981\u610f\u4e49\u3002\u7136\u800c\uff0c\u76ee\u524d\u5927\u591a\u6570\u673a\u5668\u5b66\u4e60\u6a21\u578b\u5728\u5904\u7406\u8fd9\u4e9b\u4fe1\u53f7\u65f6\uff0c\u901a\u5e38\u662f\u9488\u5bf9\u7279\u5b9a\u4efb\u52a1\u8fdb\u884c\u8bad\u7ec3\u7684\uff0c\u8fd9\u9650\u5236\u4e86\u5b83\u4eec\u5728\u4e0d\u540c\u533b\u7597\u5e94\u7528\u573a\u666f\u4e2d\u7684\u901a\u7528\u6027\u3002\u4e3a\u4e86\u514b\u670d\u8fd9\u4e00\u95ee\u9898\uff0c\u8bba\u6587\u63d0\u51fa\u91c7\u7528\u81ea\u76d1\u7763\u5b66\u4e60\u6846\u67b6\uff08SimCLR\uff09\u548cSlowfast NFNet\u4e3b\u5e72\u7f51\u7edc\u8fdb\u884c\u5bf9\u6bd4\u5b66\u4e60\uff0c\u65e8\u5728\u5f00\u53d1\u9002\u7528\u4e8e\u591a\u79cd\u5065\u5eb7\u58f0\u5b66\u4efb\u52a1\u7684\u901a\u7528\u8868\u793a\u3002<\/p>\n\n\n\n

\u4e8c\u3001\u7814\u7a76\u76ee\u6807<\/h5>\n\n\n\n

\u8bba\u6587\u7684\u6838\u5fc3\u7814\u7a76\u95ee\u9898\u662f\u5982\u4f55\u4f18\u5316Slowfast NFNet\u5728\u5065\u5eb7\u58f0\u5b66\u4fe1\u53f7\u5bf9\u6bd4\u5b66\u4e60\u4e2d\u7684\u97f3\u9891\u589e\u5f3a\u7b56\u7565\u3002\u8bba\u6587\u901a\u8fc7\u6df1\u5165\u5206\u6790\u591a\u79cd\u97f3\u9891\u589e\u5f3a\u7b56\u7565\uff0c\u53d1\u73b0\u5408\u9002\u7684\u589e\u5f3a\u7b56\u7565\u53ef\u4ee5\u663e\u8457\u63d0\u5347Slowfast NFNet\u97f3\u9891\u7f16\u7801\u5668\u5728\u591a\u79cd\u5065\u5eb7\u58f0\u5b66\u4efb\u52a1\u4e2d\u7684\u8868\u73b0\u3002<\/p>\n\n\n\n

\u4e09\u3001\u65b9\u6cd5<\/h5>\n\n\n\n

\u8bba\u6587\u7684\u7814\u7a76\u5206\u4e3a\u4e09\u4e2a\u4e3b\u8981\u9636\u6bb5\uff1a<\/p>\n\n\n\n

    \n
  1. \u786e\u5b9a\u5355\u4e2a\u589e\u5f3a\u7b56\u7565\u7684\u6700\u4f73\u53c2\u6570<\/strong>\uff1a\u901a\u8fc7\u7f51\u683c\u641c\u7d22\uff08grid search\uff09\u627e\u5230\u6bcf\u4e2a\u589e\u5f3a\u7b56\u7565\u7684\u6700\u4f18\u53c2\u6570\u3002<\/li>\n\n\n\n
  2. \u63a2\u7d22\u589e\u5f3a\u7b56\u7565\u7684\u7ec4\u5408<\/strong>\uff1a\u7814\u7a76\u5c06\u4e00\u79cd\u6216\u4e24\u79cd\u589e\u5f3a\u7b56\u7565\u8fde\u7eed\u5e94\u7528\u65f6\u7684\u6548\u679c\u3002<\/li>\n\n\n\n
  3. \u4e0e\u73b0\u6709\u6700\u5148\u8fdb\u7684\u97f3\u9891\u7f16\u7801\u5668\u8fdb\u884c\u6bd4\u8f83<\/strong>\uff1a\u5c06\u4f18\u5316\u540e\u7684Slowfast NFNet\u6a21\u578b\u4e0e\u5176\u4ed6\u97f3\u9891\u7f16\u7801\u5668\u5728\u4e94\u4e2a\u6570\u636e\u96c6\u4e0a\u768421\u4e2a\u4e8c\u5143\u5206\u7c7b\u4efb\u52a1\u4e0a\u8fdb\u884c\u6bd4\u8f83\u3002<\/li>\n<\/ol>\n\n\n\n

    \u5728\u8fd9\u4e9b\u5b9e\u9a8c\u4e2d\uff0c\u8bba\u6587\u4f7f\u7528\u4e86SimCLR\u81ea\u76d1\u7763\u5bf9\u6bd4\u5b66\u4e60\u6846\u67b6\uff0c\u5e76\u5bf9\u591a\u79cd\u97f3\u9891\u589e\u5f3a\u7b56\u7565\u8fdb\u884c\u4e86\u7cfb\u7edf\u6027\u63a2\u7d22\uff0c\u5982\u88c1\u526a\u548c\u586b\u5145\u3001\u566a\u58f0\u6dfb\u52a0\u3001\u5e03\u6717\u5e26\u901f\u7387\uff08Brownian tape speed\uff09<\/a>\u3001\u97f3\u9891\u7f29\u653e\u3001\u97f3\u9ad8\u79fb\u4f4d\u3001\u65f6\u95f4\u62c9\u4f38\u3001\u5faa\u73af\u65f6\u95f4\u79fb\u4f4d\u548cSpecAugment\u7b49\u3002\u7814\u7a76\u53d1\u73b0\uff0c\u901a\u8fc7\u7ec4\u5408\u8fd9\u4e9b\u589e\u5f3a\u7b56\u7565\uff0c\u53ef\u4ee5\u4ea7\u751f\u534f\u540c\u6548\u5e94\uff0c\u4f7f\u6a21\u578b\u8868\u73b0\u4f18\u4e8e\u5355\u72ec\u5e94\u7528\u6bcf\u79cd\u7b56\u7565\u7684\u6548\u679c\u3002<\/p>\n\n\n\n

    \"\"
    \u88c1\u526a\u548c\u586b\u5145\u3001\u566a\u58f0\u6dfb\u52a0\u3001\u5e03\u6717\u5e26\u901f\u7387\uff08Brownian tape speed\uff09<\/a>\u3001\u97f3\u9891\u7f29\u653e\u3001\u97f3\u9ad8\u79fb\u4f4d\u3001\u65f6\u95f4\u62c9\u4f38\u3001\u5faa\u73af\u65f6\u95f4\u79fb\u4f4d\u548cSpecAugment\u7b49\u97f3\u9891\u589e\u5f3a\u7b56\u7565<\/figcaption><\/figure>\n\n\n\n
    \u56db\u3001\u6570\u636e\u96c6<\/h5>\n\n\n\n

    \u8bba\u6587\u4f7f\u7528\u4e86\u4e00\u4e2a\u6765\u81eaYouTube\u7684\u975e\u8bed\u4e49\u58f0\u97f3\u6570\u636e\u96c6\uff08YT-NS\uff09\u8fdb\u884c\u8bad\u7ec3\uff0c\u5305\u542b\u7ea6142,000\u5c0f\u65f6\u7684\u97f3\u9891\u6570\u636e\uff0c\u8fd9\u4e9b\u6570\u636e\u7ecf\u8fc7\u8fc7\u6ee4\uff0c\u5305\u542b\u54b3\u55fd\u58f0\u3001\u8bb2\u8bdd\u58f0\u3001\u7b11\u58f0\u7b49\u5065\u5eb7\u76f8\u5173\u7684\u58f0\u5b66\u4fe1\u53f7\u3002\u4e3a\u4e86\u8bc4\u4f30\u6a21\u578b\u7684\u6027\u80fd\uff0c\u8bba\u6587\u4f7f\u7528\u4e86\u4e94\u4e2a\u516c\u5f00\u6570\u636e\u96c6\uff0c\u5305\u62ecFSD50K\u3001Flusense\u3001PSG\u3001CoughVID\u548cCoswara\u3002<\/p>\n\n\n\n

    \u4e94\u3001\u5b9e\u9a8c\u7ed3\u679c<\/h5>\n\n\n\n

    \u7814\u7a76\u8868\u660e\uff0c\u7ec4\u5408\u4f7f\u7528\u589e\u5f3a\u7b56\u7565\u5f80\u5f80\u6bd4\u5355\u72ec\u4f7f\u7528\u67d0\u4e00\u589e\u5f3a\u7b56\u7565\u6548\u679c\u66f4\u597d<\/strong>\u3002\u5177\u4f53\u6765\u8bf4\uff0c\u7814\u7a76\u53d1\u73b0\u201c\u5faa\u73af\u65f6\u95f4\u79fb\u4f4d\u201d\u7ed3\u5408\u201c\u65f6\u95f4\u62c9\u4f38\u201d\u662f\u6700\u6709\u6548\u7684\u4e24\u6b65\u589e\u5f3a\u7b56\u7565<\/strong>\uff0c\u5c3d\u7ba1\u5355\u72ec\u4f7f\u7528\u65f6\u8fd9\u4e24\u79cd\u7b56\u7565\u7684\u6548\u679c\u4e0d\u5982SpecAugment\uff0c\u4f46\u5b83\u4eec\u7ec4\u5408\u540e\u8868\u73b0\u51fa\u663e\u8457\u7684\u534f\u540c\u6548\u5e94\u3002\u6700\u7ec8\u4f18\u5316\u540e\u7684SimCLR\u6a21\u578b\u5728\u591a\u4e2a\u4e0b\u6e38\u4efb\u52a1\u4e2d\u8868\u73b0\u51fa\u4e0e\u6700\u5148\u8fdb\u7684\u97f3\u9891\u7f16\u7801\u5668\u76f8\u5f53\u751a\u81f3\u66f4\u4f18\u7684\u6027\u80fd\u3002<\/p>\n\n\n\n

    \u516d\u3001\u7ed3\u8bba<\/h5>\n\n\n\n

    \u8bba\u6587\u901a\u8fc7\u7cfb\u7edf\u5730\u63a2\u8ba8\u5065\u5eb7\u58f0\u5b66\u9886\u57df\u7684\u97f3\u9891\u589e\u5f3a\u7b56\u7565\uff0c\u63ed\u793a\u4e86\u5faa\u73af\u65f6\u95f4\u79fb\u4f4d\u548c\u65f6\u95f4\u62c9\u4f38\u7ec4\u5408\u5728\u589e\u5f3a\u6a21\u578b\u6cdb\u5316\u80fd\u529b\u65b9\u9762\u7684\u6f5c\u529b\u3002\u8fd9\u4e9b\u7814\u7a76\u7ed3\u679c\u8868\u660e\uff0c\u9002\u5f53\u7684\u589e\u5f3a\u7b56\u7565\u7ec4\u5408\u53ef\u4ee5\u663e\u8457\u63d0\u5347\u5bf9\u6bd4\u5b66\u4e60\u5728\u5065\u5eb7\u76f8\u5173\u58f0\u5b66\u4fe1\u53f7\u4e0a\u7684\u8868\u73b0\u3002\u7136\u800c\uff0c\u4f5c\u8005\u4e5f\u6307\u51fa\u4e86\u4e00\u4e9b\u9650\u5236\uff0c\u5982\u5b9e\u9a8c\u53ea\u5728Slowfast NFNet\u67b6\u6784\u4e0a\u8fdb\u884c\uff0c\u672a\u6765\u7814\u7a76\u53ef\u4ee5\u5c1d\u8bd5\u5176\u4ed6\u67b6\u6784\u548c\u589e\u5f3a\u7b56\u7565\u3002<\/p>\n\n\n\n

    \u4e03\u3001\u9650\u5236\u4e0e\u672a\u6765\u5de5\u4f5c<\/h5>\n\n\n\n

    \u7814\u7a76\u4fdd\u7559\u4e86\u6d4b\u8bd5\u96c6\uff0c\u4ee5\u4fbf\u8fdb\u884c\u6301\u7eed\u7684\u6a21\u578b\u5f00\u53d1\uff0c\u8fd9\u610f\u5473\u7740\u5b9e\u9a8c\u7ed3\u679c\u53ef\u80fd\u5177\u6709\u4e00\u5b9a\u7684\u4e50\u89c2\u6027\u3002\u6b64\u5916\uff0c\u7814\u7a76\u4ec5\u5728Slowfast NFNet\u67b6\u6784\u4e0a\u8fdb\u884c\u4e86\u5206\u6790\uff0c\u672a\u6765\u53ef\u4ee5\u5728\u4e0d\u540c\u7684\u67b6\u6784\u548c\u589e\u5f3a\u7b56\u7565\u4e0a\u8fdb\u4e00\u6b65\u9a8c\u8bc1\u7ed3\u679c\u3002\u4f5c\u8005\u8fd8\u5efa\u8bae\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u5f15\u5165\u5065\u5eb7\u4fe1\u53f7\u7c7b\u578b\u6807\u7b7e\uff0c\u53ef\u80fd\u8fdb\u4e00\u6b65\u63d0\u9ad8\u5b66\u4e60\u5230\u7684\u8868\u793a\u7684\u8d28\u91cf\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"

    \u8bba\u6587Optimizing Audio Augmentations for Contrastive Learni […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[8,36],"tags":[39,99,45,66],"class_list":["post-4492","post","type-post","status-publish","format-standard","hentry","category-tech","category-36","tag-ai","tag-health","tag-medicine","tag-google"],"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"NullThought","author_link":"https:\/\/nullthought.net\/?author=1"},"rttpg_comment":0,"rttpg_category":"Tech<\/a> \u79d1\u5b66<\/a>","rttpg_excerpt":"\u8bba\u6587Optimizing Audio Augmentations for Contrastive Learni…","_links":{"self":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/4492","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=4492"}],"version-history":[{"count":2,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/4492\/revisions"}],"predecessor-version":[{"id":4509,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/4492\/revisions\/4509"}],"wp:attachment":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4492"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4492"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4492"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}