{"id":11621,"date":"2020-08-23T08:13:15","date_gmt":"2020-08-23T12:13:15","guid":{"rendered":"http:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/?p=11621"},"modified":"2020-08-23T08:13:15","modified_gmt":"2020-08-23T12:13:15","slug":"dealing-with-overconfidence-in-neural-networks-bayesian-approach","status":"publish","type":"post","link":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/dealing-with-overconfidence-in-neural-networks-bayesian-approach\/11621\/","title":{"rendered":"Dealing with Overconfidence in Neural Networks: Bayesian Approach"},"content":{"rendered":"Originally published in Jonathan Ramkissoon Blog, July 29, 2020. I trained a multi-class classifier on images of cats, dogs and wild animals and passed an image of myself, it\u2019s 98% confident I\u2019m a dog. The problem isn\u2019t that I passed an inappropriate image, because models in the real world are passed all sorts of garbage. It\u2019s that the model is overconfident about an image far away from the training data. Instead we expect a more uniform distribution over the classes. The overconfidence makes it difficult to post-process model output (setting a threshold on predictions, etc.), which means it <a href=\"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/dealing-with-overconfidence-in-neural-networks-bayesian-approach\/11621\/\" class=\"more-link\">(more&hellip;)<\/a>","protected":false},"excerpt":{"rendered":"<p>Originally published in Jonathan Ramkissoon Blog, July 29, 2020. I trained a multi-class classifier on images of cats, dogs and wild animals and passed an image of myself, it\u2019s 98% confident I\u2019m a dog. The problem isn\u2019t that I passed an inappropriate image, because models in the real world are passed all sorts of garbage. [&hellip;]<\/p>\n","protected":false},"author":72,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":"","_links_to":"","_links_to_target":""},"categories":[11,48],"tags":[59,230,791,243,1119],"class_list":["post-11621","post","type-post","status-publish","format-standard","hentry","category-industry-news","category-left-hand","tag-analytics","tag-data-analytics","tag-deep-learning","tag-machine-learning","tag-neural-networks"],"_links":{"self":[{"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/posts\/11621","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/users\/72"}],"replies":[{"embeddable":true,"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/comments?post=11621"}],"version-history":[{"count":1,"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/posts\/11621\/revisions"}],"predecessor-version":[{"id":11622,"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/posts\/11621\/revisions\/11622"}],"wp:attachment":[{"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/media?parent=11621"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/categories?post=11621"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/wp-json\/wp\/v2\/tags?post=11621"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}