{"id":585736,"date":"2019-04-10T00:00:34","date_gmt":"2019-04-10T07:00:34","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-research-item&#038;p=585736"},"modified":"2019-05-08T16:38:39","modified_gmt":"2019-05-08T23:38:39","slug":"how-does-batch-normalization-help-optimization","status":"publish","type":"msr-video","link":"https:\/\/www.microsoft.com\/en-us\/research\/video\/how-does-batch-normalization-help-optimization\/","title":{"rendered":"How does Batch Normalization Help Optimization?"},"content":{"rendered":"<p>Batch normalization (BatchNorm) is a widely adopted technique that enables faster and more stable training of deep neural networks. However, despite its pervasiveness, the exact reasons for BatchNorm&#8217;s effectiveness are still poorly understood.<\/p>\n<p>In this talk, we take a closer look at the underpinnings of the BatchNorm&#8217;s success. In particular, we examine the popular belief that the root of BatchNorm&#8217;s effectiveness is due to reduction of an effect called internal covariate shift (ICS). We then explore the connection between BatchNorm, ICS, and the optimization landscape of deep neural networks.<\/p>\n<p><a href=\"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2019\/05\/How-does-Batch-Normalization-Help-Optimization-slides.pdf\" target=\"_blank\" rel=\"noopener noreferrer\">[SLIDES]<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Batch normalization (BatchNorm) is a widely adopted technique that enables faster and more stable training of deep neural networks. However, despite its pervasiveness, the exact reasons for BatchNorm&#8217;s effectiveness are still poorly understood. In this talk, we take a closer look at the underpinnings of the BatchNorm&#8217;s success. In particular, we examine the popular belief [&hellip;]<\/p>\n","protected":false},"featured_media":585754,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr_hide_image_in_river":0,"footnotes":""},"research-area":[13556],"msr-video-type":[206954],"msr-locale":[268875],"msr-post-option":[],"msr-session-type":[],"msr-impact-theme":[],"msr-pillar":[],"msr-episode":[],"msr-research-theme":[],"class_list":["post-585736","msr-video","type-msr-video","status-publish","has-post-thumbnail","hentry","msr-research-area-artificial-intelligence","msr-video-type-microsoft-research-talks","msr-locale-en_us"],"msr_download_urls":"","msr_external_url":"https:\/\/youtu.be\/EvAVCxZJN2U","msr_secondary_video_url":"","msr_video_file":"","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video\/585736","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-video"}],"version-history":[{"count":3,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video\/585736\/revisions"}],"predecessor-version":[{"id":585748,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video\/585736\/revisions\/585748"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media\/585754"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=585736"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=585736"},{"taxonomy":"msr-video-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video-type?post=585736"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=585736"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=585736"},{"taxonomy":"msr-session-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-session-type?post=585736"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=585736"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=585736"},{"taxonomy":"msr-episode","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-episode?post=585736"},{"taxonomy":"msr-research-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-theme?post=585736"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}