{"id":4110,"date":"2024-07-16T19:41:47","date_gmt":"2024-07-16T11:41:47","guid":{"rendered":"https:\/\/nullthought.net\/?p=4110"},"modified":"2024-10-16T10:45:17","modified_gmt":"2024-10-16T02:45:17","slug":"gbrl-gradient-boosting-reinforcement-learning-%e6%a2%af%e5%ba%a6%e5%a2%9e%e5%bc%ba%e5%bc%ba%e5%8c%96%e5%ad%a6%e4%b9%a0","status":"publish","type":"post","link":"https:\/\/nullthought.net\/?p=4110","title":{"rendered":"GBRL, Gradient Boosting Reinforcement Learning, \u68af\u5ea6\u589e\u5f3a\u5f3a\u5316\u5b66\u4e60"},"content":{"rendered":"\n<h4 class=\"wp-block-heading\">\u6458\u8981<\/h4>\n\n\n\n<p>\u8bba\u6587<strong><em><a href=\"https:\/\/arxiv.org\/abs\/2407.08250\" target=\"_blank\" rel=\"noreferrer noopener\">Gradient Boosting Reinforcement Learning<\/a><\/em><\/strong>\u4ecb\u7ecd\u4e86\u68af\u5ea6\u589e\u5f3a\u5f3a\u5316\u5b66\u4e60\uff08GBRL\uff09\uff0c\u5b83\u5c06\u68af\u5ea6\u589e\u5f3a\u6811\uff08GBT\uff09\u4e0e\u5f3a\u5316\u5b66\u4e60\uff08RL\uff09\u7ed3\u5408\u3002\u4f5c\u8005\u6307\u51fa\uff0c\u5c3d\u7ba1\u795e\u7ecf\u7f51\u7edc\u5728\u5404\u79cd\u4efb\u52a1\u4e2d\u53d6\u5f97\u4e86\u663e\u8457\u6210\u679c\uff0c\u4f46\u5b83\u4eec\u901a\u5e38\u7f3a\u4e4f\u53ef\u89e3\u91ca\u6027\uff0c\u4e0d\u9002\u5408\u5904\u7406\u5206\u7c7b\u6570\u636e\u6216\u5728\u8fb9\u7f18\u8bbe\u5907\u4e0a\u5b9e\u73b0\u8f7b\u91cf\u7ea7\u5e94\u7528\u3002\u800cGBT\u81ea\u7136\u9002\u5408\u8fd9\u4e9b\u8981\u6c42\u3002GBRL\u5c06GBT\u7684\u4f18\u52bf\u6269\u5c55\u5230RL\u9886\u57df\uff0c\u5728\u5404\u79cd\u4efb\u52a1\u4e2d\u5b9e\u73b0\u4e86\u7ade\u4e89\u6027\u6027\u80fd\uff0c\u5c24\u5176\u662f\u5728\u5177\u6709\u7ed3\u6784\u5316\u6216\u5206\u7c7b\u7279\u5f81\u7684\u9886\u57df\u3002\u6b64\u5916\uff0c\u4f5c\u8005\u8fd8\u63d0\u4f9b\u4e86\u9ad8\u6027\u80fd\u7684GPU\u52a0\u901f\u5b9e\u73b0\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u4f5c\u8005\u4e3a\u6765\u81eaNvidia\u7684Benjamin Fuhrer, Chen Tessler\u548cGal Dalal\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u5185\u5bb9\u6982\u8981\u5982\u4e0b\uff1a<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u4e00\u3001\u5f15\u8a00<\/h4>\n\n\n\n<p>\u5f15\u8a00\u5f3a\u8c03\u4e86\u9700\u8981\u53ef\u89e3\u91ca\u5e76\u80fd\u591f\u6709\u6548\u5904\u7406\u5b9e\u9645\u4efb\u52a1\u4e2d\u5e38\u89c1\u7684\u7ed3\u6784\u5316\u548c\u5206\u7c7b\u6570\u636e\u7684RL\u89e3\u51b3\u65b9\u6848\u3002\u4f5c\u8005\u6307\u51fa\uff0c\u5f53\u524d\u57fa\u4e8e\u795e\u7ecf\u7f51\u7edc\u7684\u89e3\u51b3\u65b9\u6848\u5728\u8fd9\u4e9b\u8981\u6c42\u4e0a\u8868\u73b0\u4e0d\u4f73\u3002\u4ed6\u4eec\u63d0\u51fa\u4e86GBRL\uff0c\u4ee5\u5229\u7528GBT\u5728RL\u573a\u666f\u4e2d\u7684\u4f18\u52bf\uff0c\u4ece\u800c\u586b\u8865\u8fd9\u4e00\u5173\u952e\u7a7a\u767d\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u4e8c\u3001\u8d21\u732e<\/h4>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>GBT\u7528\u4e8eRL<\/strong>\uff1a\u5c55\u793a\u4e86GBT\u5728RL\u4e2d\u4f5c\u4e3a\u51fd\u6570\u8fd1\u4f3c\u5668\u7684\u53ef\u884c\u6027\uff0c\u901a\u8fc7\u5b9e\u73b0\u5404\u79cdactor-critic\u7b97\u6cd5\u3002<\/li>\n\n\n\n<li><strong>\u57fa\u4e8e\u6811\u7684Actor-Critic\u67b6\u6784<\/strong>\uff1a\u4ecb\u7ecd\u4e86\u4e00\u79cd\u7528\u4e8e\u7b56\u7565\u548c\u4ef7\u503c\u51fd\u6570\u7684\u5171\u4eab\u67b6\u6784\uff0c\u51cf\u5c11\u4e86\u5185\u5b58\u548c\u8ba1\u7b97\u9700\u6c42\u3002<\/li>\n\n\n\n<li><strong>\u73b0\u4ee3GBT RL\u5e93<\/strong>\uff1a\u63d0\u4f9b\u4e86\u4e00\u4e2a\u57fa\u4e8eCUDA\u7684GPU\u52a0\u901fGBT\u6846\u67b6\uff0c\u53ef\u4e0e\u6d41\u884c\u7684RL\u5e93\u65e0\u7f1d\u96c6\u6210\u3002<\/li>\n<\/ol>\n\n\n\n<h4 class=\"wp-block-heading\">\u4e09\u3001\u76f8\u5173\u5de5\u4f5c<\/h4>\n\n\n\n<p>\u672c\u6587\u8ba8\u8bba\u4e86GBT\u7684\u76f8\u5173\u5de5\u4f5c\u53ca\u5176\u5728RL\u4e2d\u7684\u6709\u9650\u5e94\u7528\u3002\u5b83\u56de\u987e\u4e86GBT\u5728\u56de\u5f52\u3001\u5206\u7c7b\u548c\u6392\u5e8f\u65b9\u9762\u7684\u8fdb\u5c55\uff0c\u5e76\u5c06\u5176\u4e0e\u6240\u63d0\u51fa\u7684GBRL\u6846\u67b6\u8fdb\u884c\u4e86\u6bd4\u8f83\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u56db\u3001\u9884\u5907\u77e5\u8bc6<\/h4>\n\n\n\n<p>\u672c\u8282\u4ecb\u7ecd\u4e86\u9a6c\u5c14\u53ef\u592b\u51b3\u7b56\u8fc7\u7a0b\uff08MDP\uff09\u548cRL\u4e2d\u7684actor-critic\u65b9\u6cd5\u7684\u57fa\u7840\u77e5\u8bc6\u3002\u8fd8\u89e3\u91ca\u4e86GBT\u53ca\u5176\u5728\u51fd\u6570\u68af\u5ea6\u4e0b\u964d\u4e2d\u7684\u4f7f\u7528\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u4e94\u3001\u68af\u5ea6\u589e\u5f3a\u5f3a\u5316\u5b66\u4e60<\/h4>\n\n\n\n<p>\u6838\u5fc3\u90e8\u5206\u63cf\u8ff0\u4e86GBRL\u5982\u4f55\u5c06GBT\u9002\u5e94\u4e8eRL\u4e2d\u7684actor-critic\u7b97\u6cd5\u3002\u8be5\u65b9\u6cd5\u6d89\u53ca\u901a\u8fc7\u8fed\u4ee3\u5730\u589e\u957f\u6811\u7684\u96c6\u5408\u6765\u4f18\u5316RL\u76ee\u6807\u3002\u672c\u6587\u8fd8\u89e3\u51b3\u4e86RL\u4e2d\u7684\u72ec\u7279\u6311\u6218\uff0c\u5982\u975e\u5e73\u7a33\u72b6\u6001\u5206\u5e03\u548c\u5728\u7ebf\u5b66\u4e60\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u516d\u3001\u5b9e\u9a8c<\/h4>\n\n\n\n<p>\u5b9e\u9a8c\u65e8\u5728\u8bc4\u4f30GBRL\u5728\u5404\u79cdRL\u4efb\u52a1\u4e2d\u7684\u8868\u73b0\uff0c\u5e76\u4e0e\u57fa\u4e8e\u795e\u7ecf\u7f51\u7edc\u7684\u65b9\u6cd5\u8fdb\u884c\u6bd4\u8f83\u3002\u6d4b\u8bd5\u73af\u5883\u5305\u62ec\u7ecf\u5178\u63a7\u5236\u4efb\u52a1\u3001\u9ad8\u7ef4\u77e2\u91cf\u5316\u95ee\u9898\u548c\u5206\u7c7b\u4efb\u52a1\u3002\u7ed3\u679c\u8868\u660e\uff0cGBRL\u5728\u7ed3\u6784\u5316\u548c\u5206\u7c7b\u73af\u5883\u4e2d\u8868\u73b0\u826f\u597d\uff0c\u4f46\u5728Atari-RAM\u7b49\u975e\u7ed3\u6784\u5316\u4efb\u52a1\u4e2d\u8868\u73b0\u8f83\u5dee\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u4e03\u3001\u7ed3\u679c\u603b\u7ed3<\/h4>\n\n\n\n<ul class=\"wp-block-list\">\n<li>GBRL\u5728\u7ed3\u6784\u5316\u4efb\u52a1\u4e2d\u4e0e\u795e\u7ecf\u7f51\u7edc\u5177\u6709\u7ade\u4e89\u529b\u3002<\/li>\n\n\n\n<li>\u7ed3\u5408PPO\uff08\u8fd1\u7aef\u7b56\u7565\u4f18\u5316\uff09\u7684GBRL\u8868\u73b0\u5c24\u4e3a\u51fa\u8272\u3002<\/li>\n\n\n\n<li>\u50cfXGBoost\u548cCatBoost\u8fd9\u6837\u7684\u6807\u51c6GBT\u5e93\u5728RL\u4efb\u52a1\u4e2d\u4e0d\u5982GBRL\u9ad8\u6548\u3002<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">\u516b\u3001\u7ed3\u8bba<\/h4>\n\n\n\n<p>\u672c\u6587\u603b\u7ed3\u9053\uff0cGBRL\u6210\u529f\u5730\u5c06GBT\u6574\u5408\u5230RL\u5de5\u5177\u4e2d\uff0c\u63d0\u4f9b\u4e86\u9002\u7528\u4e8e\u7ed3\u6784\u5316\u6570\u636e\u73af\u5883\u7684\u53ef\u89e3\u91ca\u4e14\u9ad8\u6548\u7684\u89e3\u51b3\u65b9\u6848\u3002\u672a\u6765\u7684\u5de5\u4f5c\u53ef\u80fd\u96c6\u4e2d\u5728\u89e3\u51b3\u5982\u96c6\u6210\u5927\u5c0f\u548c\u8ba1\u7b97\u6548\u7387\u7b49\u9650\u5236\u4e0a\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u4e5d\u3001\u9650\u5236\u548c\u672a\u6765\u65b9\u5411<\/h4>\n\n\n\n<p>\u4f5c\u8005\u627f\u8ba4\u5b58\u5728\u96c6\u6210\u5927\u5c0f\u589e\u957f\u548c\u8ba1\u7b97\u6548\u7387\u7b49\u6311\u6218\u3002\u4ed6\u4eec\u5efa\u8bae\u63a2\u7d22\u6811\u4fee\u526a\u3001\u96c6\u6210\u538b\u7f29\u548c\u52a8\u6001\u96c6\u6210\u7ba1\u7406\u7b49\u6f5c\u5728\u7684\u672a\u6765\u7814\u7a76\u9886\u57df\u3002\u6b64\u5916\uff0c\u5c06GBT\u4e0e\u9700\u8981\u53ef\u5fae\u5206Q\u51fd\u6570\u7684\u7b97\u6cd5\u96c6\u6210\u4ecd\u7136\u662f\u4e00\u4e2a\u5f00\u653e\u7684\u6311\u6218\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">\u5341\u3001\u603b\u7ed3<\/h3>\n\n\n\n<p>\u8bba\u6587\u4ecb\u7ecd\u4e86\u4e00\u79cd\u901a\u8fc7\u5c06\u68af\u5ea6\u589e\u5f3a\u6811\u4e0e\u5f3a\u5316\u5b66\u4e60\u76f8\u7ed3\u5408\u7684\u521b\u65b0\u65b9\u6cd5\u3002\u8be5\u65b9\u6cd5\u89e3\u51b3\u4e86\u795e\u7ecf\u7f51\u7edc\u5728\u53ef\u89e3\u91ca\u6027\u548c\u5904\u7406\u7ed3\u6784\u5316\/\u5206\u7c7b\u6570\u636e\u65b9\u9762\u7684\u5c40\u9650\u6027\uff0c\u63d0\u4f9b\u4e86\u4e00\u79cd\u5728\u7279\u5b9a\u9886\u57df\u5177\u6709\u7ade\u4e89\u529b\u7684\u66ff\u4ee3\u65b9\u6848\u3002\u63d0\u4f9b\u7684GPU\u52a0\u901f\u5b9e\u73b0\u548c\u8be6\u7ec6\u7684\u5b9e\u9a8c\u7a81\u51fa\u4e86GBRL\u5728\u5404\u79cdRL\u4efb\u52a1\u4e2d\u7684\u5b9e\u9645\u5e94\u7528\u6027\u548c\u4f18\u52bf\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>GBRL on GitHub: <a href=\"https:\/\/github.com\/NVlabs\/gbrl\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/github.com\/NVlabs\/gbrl<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u6458\u8981 \u8bba\u6587Gradient Boosting Reinforcement Learning\u4ecb\u7ecd\u4e86\u68af\u5ea6\u589e\u5f3a\u5f3a\u5316\u5b66 [&hellip;]<\/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":[39,84],"class_list":["post-4110","post","type-post","status-publish","format-standard","hentry","category-tech","category-36","tag-ai","tag-nvidia"],"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"NullThought","author_link":"https:\/\/nullthought.net\/?author=1"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/nullthought.net\/?cat=8\" rel=\"category\">Tech<\/a> <a href=\"https:\/\/nullthought.net\/?cat=36\" rel=\"category\">\u79d1\u5b66<\/a>","rttpg_excerpt":"\u6458\u8981 \u8bba\u6587Gradient Boosting Reinforcement Learning\u4ecb\u7ecd\u4e86\u68af\u5ea6\u589e\u5f3a\u5f3a\u5316\u5b66&hellip;","_links":{"self":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/4110","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=4110"}],"version-history":[{"count":1,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/4110\/revisions"}],"predecessor-version":[{"id":4111,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/4110\/revisions\/4111"}],"wp:attachment":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4110"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4110"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4110"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}