{"id":656649,"date":"2019-06-07T09:45:49","date_gmt":"2019-06-07T16:45:49","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-research-item&#038;p=656649"},"modified":"2020-05-05T13:55:26","modified_gmt":"2020-05-05T20:55:26","slug":"fast-and-flexible-multi-task-classification-using-conditional-neural-adaptive-processes","status":"publish","type":"msr-video","link":"https:\/\/www.microsoft.com\/en-us\/research\/video\/fast-and-flexible-multi-task-classification-using-conditional-neural-adaptive-processes\/","title":{"rendered":"Fast and Flexible Multi-Task Classification Using Conditional Neural Adaptive Processes"},"content":{"rendered":"<p>This talk will describe our recent work on designing image classification systems that, after an initial multi-task training phase, can automatically adapt to new tasks encountered at test time. I will introduce an approach that relates to existing approaches to meta-learning and so-called conditional neural processes, generalising them to the multi-task classification setting. The resulting approach, called Conditional Neural Adaptive Processes (CNAPS), comprises a classifier whose parameters are modulated by an adaptation network that takes the current task&#8217;s dataset as input. I will show that CNAPS achieves state-of-the-art results on the challenging Meta-Dataset few-shot learning benchmark indicating high-quality transfer-learning which is robust, avoiding both over-fitting in low-shot regimes and under-fitting in high-shot regimes. Timing experiments reveal that CNAPS is computationally efficient at test-time as it does not involve gradient based adaptation. Finally, I will show that trained models are immediately deployable to continual learning and active learning where they can outperform existing approaches that do not leverage transfer learning.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>This talk will describe our recent work on designing image classification systems that, after an initial multi-task training phase, can automatically adapt to new tasks encountered at test time. I will introduce an approach that relates to existing approaches to meta-learning and so-called conditional neural processes, generalising them to the multi-task classification setting. The resulting [&hellip;]<\/p>\n","protected":false},"featured_media":656652,"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":[],"msr-locale":[268875],"msr-post-option":[],"msr-session-type":[],"msr-impact-theme":[],"msr-pillar":[],"msr-episode":[],"msr-research-theme":[],"class_list":["post-656649","msr-video","type-msr-video","status-publish","has-post-thumbnail","hentry","msr-research-area-artificial-intelligence","msr-locale-en_us"],"msr_download_urls":"","msr_external_url":"https:\/\/youtu.be\/apayUKSExmU","msr_secondary_video_url":"","msr_video_file":"","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video\/656649","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":1,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video\/656649\/revisions"}],"predecessor-version":[{"id":656655,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video\/656649\/revisions\/656655"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media\/656652"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=656649"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=656649"},{"taxonomy":"msr-video-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video-type?post=656649"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=656649"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=656649"},{"taxonomy":"msr-session-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-session-type?post=656649"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=656649"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=656649"},{"taxonomy":"msr-episode","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-episode?post=656649"},{"taxonomy":"msr-research-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-theme?post=656649"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}