{"id":73584,"date":"2017-10-30T09:02:30","date_gmt":"2017-10-30T16:02:30","guid":{"rendered":"https:\/\/lifeboat.com\/blog\/2017\/10\/how-do-you-turn-a-dog-into-a-car-change-a-single-pixel"},"modified":"2017-10-30T09:02:30","modified_gmt":"2017-10-30T16:02:30","slug":"how-do-you-turn-a-dog-into-a-car-change-a-single-pixel","status":"publish","type":"post","link":"https:\/\/lifeboat.com\/blog\/2017\/10\/how-do-you-turn-a-dog-into-a-car-change-a-single-pixel","title":{"rendered":"How Do You Turn a Dog into a Car? Change a Single Pixel"},"content":{"rendered":"<p><a class=\"aligncenter blog-photo\" href=\"https:\/\/lifeboat.com\/blog.images\/how-do-you-turn-a-dog-into-a-car-change-a-single-pixel.jpg\"><\/a><\/p>\n<p>Thank a new approach to spoofing image recognition AIs, developed by a team from Kyushu University in Japan, for that joke.<\/p>\n<p>Trying to catch out AIs is a <a href=\"https:\/\/www.technologyreview.com\/s\/608381\/ai-shouldnt-believe-everything-it-hears\/\" target=\"_blank\">popular pastime for many researchers<\/a>, and we\u2019ve reported machine-learning spoofs in the past. The general approach is to <a href=\"https:\/\/www.technologyreview.com\/s\/601955\/machine-visions-achilles-heel-revealed-by-google-brain-researchers\/\" target=\"_blank\">add features to images that will incorrectly trigger a neural network<\/a> and have it identify what it sees as something else entirely.<\/p>\n<p>The new research, <a href=\"https:\/\/arxiv.org\/abs\/1710.08864\" target=\"_blank\">published on the arXiv<\/a>, describes an algorithm that can efficiently identify the best pixels to alter in order to confuse an AI into mislabeling a picture. By changing just one pixel in a 1,024-pixel image, the software can trick an AI about 74 percent of the time. That figure rises to around 87 percent if five pixels are tweaked.<\/p>\n<p><!-- Link: <a href=\"https:\/\/www.technologyreview.com\/the-download\/609263\/how-do-you-turn-a-dog-into-a-car-change-a-single-pixel\/?utm_campaign=Owned+Social&utm_source=Owned+Social&utm_medium=Facebook\">https:\/\/www.technologyreview.com\/the-download\/609263\/how-do-...m=Facebook<\/a> --><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Thank a new approach to spoofing image recognition AIs, developed by a team from Kyushu University in Japan, for that joke. Trying to catch out AIs is a popular pastime for many researchers, and we\u2019ve reported machine-learning spoofs in the past. The general approach is to add features to images that will incorrectly trigger a [\u2026]<\/p>\n","protected":false},"author":368,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[23,41,6,1491],"tags":[],"class_list":["post-73584","post","type-post","status-publish","format-standard","hentry","category-humor","category-information-science","category-robotics-ai","category-transportation"],"_links":{"self":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts\/73584","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/users\/368"}],"replies":[{"embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/comments?post=73584"}],"version-history":[{"count":0,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts\/73584\/revisions"}],"wp:attachment":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/media?parent=73584"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/categories?post=73584"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/tags?post=73584"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}