{"id":45687,"date":"2020-12-01T23:49:47","date_gmt":"2020-12-02T04:49:47","guid":{"rendered":"https:\/\/www.opensource.im\/uncategorized\/the-way-we-train-ai-is-fundamentally-flawed-machine-learning-times-the-predictive-analytics-times.php"},"modified":"2020-12-01T23:49:47","modified_gmt":"2020-12-02T04:49:47","slug":"the-way-we-train-ai-is-fundamentally-flawed-machine-learning-times-the-predictive-analytics-times","status":"publish","type":"post","link":"https:\/\/euvolution.com\/open-source-convergence\/machine-learning\/the-way-we-train-ai-is-fundamentally-flawed-machine-learning-times-the-predictive-analytics-times.php","title":{"rendered":"The Way We Train AI is Fundamentally Flawed  Machine Learning Times &#8211; The Predictive Analytics Times"},"content":{"rendered":"<p><p>Its no secret that machine-learning models tuned and tweaked to near-perfect performance in the lab often fail in real settings. This is typically put down to a mismatch between the data the AI was trained and tested on and the data it encounters in the world, a problem known as data shift. For example, an AI trained to spot signs of disease in high-quality medical images will struggle with blurry or cropped images captured by a cheap camera in a busy clinic.<\/p>\n<p>Now a group of 40 researchers across seven different teams at Google have identified another major cause for the common failure of machine-learning models. Called underspecification, it could be an even bigger problem than data shift. We are asking more of machine-learning models than we are able to guarantee with our current approach, says Alex DAmour, who led the study.<\/p>\n<p><!-- Auto Generated --><\/p>\n<p>Visit link:<br \/>\n<a target=\"_blank\" href=\"https:\/\/www.predictiveanalyticsworld.com\/machinelearningtimes\/the-way-we-train-ai-is-fundamentally-flawed\/11886\/\" title=\"The Way We Train AI is Fundamentally Flawed  Machine Learning Times - The Predictive Analytics Times\" rel=\"noopener noreferrer\">The Way We Train AI is Fundamentally Flawed  Machine Learning Times - The Predictive Analytics Times<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p> Its no secret that machine-learning models tuned and tweaked to near-perfect performance in the lab often fail in real settings. This is typically put down to a mismatch between the data the AI was trained and tested on and the data it encounters in the world, a problem known as data shift. <\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[27373],"tags":[],"class_list":["post-45687","post","type-post","status-publish","format-standard","hentry","category-machine-learning"],"_links":{"self":[{"href":"https:\/\/euvolution.com\/open-source-convergence\/wp-json\/wp\/v2\/posts\/45687"}],"collection":[{"href":"https:\/\/euvolution.com\/open-source-convergence\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/euvolution.com\/open-source-convergence\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/euvolution.com\/open-source-convergence\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/euvolution.com\/open-source-convergence\/wp-json\/wp\/v2\/comments?post=45687"}],"version-history":[{"count":0,"href":"https:\/\/euvolution.com\/open-source-convergence\/wp-json\/wp\/v2\/posts\/45687\/revisions"}],"wp:attachment":[{"href":"https:\/\/euvolution.com\/open-source-convergence\/wp-json\/wp\/v2\/media?parent=45687"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/euvolution.com\/open-source-convergence\/wp-json\/wp\/v2\/categories?post=45687"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/euvolution.com\/open-source-convergence\/wp-json\/wp\/v2\/tags?post=45687"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}