{"id":18019,"date":"2020-02-26T09:32:00","date_gmt":"2020-02-26T14:32:00","guid":{"rendered":"https:\/\/michigan.it.umich.edu\/news\/?p=18019"},"modified":"2024-07-08T06:05:05","modified_gmt":"2024-07-08T10:05:05","slug":"emotion-recognition-has-a-privacy-problem-heres-how-to-fix-it","status":"publish","type":"post","link":"https:\/\/michigan.it.umich.edu\/news\/2020\/02\/26\/emotion-recognition-has-a-privacy-problem-heres-how-to-fix-it\/","title":{"rendered":"Emotion recognition has a privacy problem \u2013 here\u2019s how to fix it"},"content":{"rendered":"\n<div class=\"wp-block-image\"><figure class=\"aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"700\" height=\"466\" src=\"https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-700x466.jpg\" alt=\"Young blonde woman with black box over eyes talking on smartphone\" class=\"wp-image-18021\"\/><figcaption><a href=\"https:\/\/freerangestock.com\/photos\/114644\/young-blonde-woman-talking-on-mobile-phone.html\/\">(Michigan Engineer via Freerange, CC0)<\/a>\n\n<\/figcaption><\/figure><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">With devices listening everywhere you go, privacy concerns are endemic to advancing technology. For example, a virtual assistant that can learn to adapt to a user&#8217;s mood creates more useful, human-like interactions. But what if the audio powering these insights stores the user&#8217;s gender and demographic information on company servers \u2013 leaving the user open to identification by the company or, worse, any malicious eavesdroppers?<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Research by CSE PhD student <strong>Mimansa Jaiswal<\/strong> and professor <strong>Emily Mower Provost<\/strong> proposes <a href=\"https:\/\/news.engin.umich.edu\/2020\/02\/emotion-recognition-has-a-privacy-problem-heres-how-to-fix-it\/\">more secure technologies built on machine learning<\/a> (ML). Through the use of adversarial ML, they\u2019ve demonstrated the ability to \u201cunlearn\u201d these sensitive identifiers from audio before it\u2019s stored, and instead use stripped-down representations of the speaker to train emotion recognition models.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u201cWe find that the performance is either maintained, or there is a slight decrease in performance for some setups,\u201d Jaiswal and Mower Provost write about their work. Jaiswal hopes to use these findings to make machine learning research safer and more secure for users in the real world.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>With devices listening everywhere you go, privacy concerns are endemic to advancing technology. For example, a virtual assistant that can learn to adapt to a user&#8217;s mood creates more useful, human-like interactions. But what if the audio powering these insights stores the user&#8217;s gender and demographic information on company servers \u2013 leaving the user open to identification by\u2026 <span class=\"read-more\"><a href=\"https:\/\/michigan.it.umich.edu\/news\/2020\/02\/26\/emotion-recognition-has-a-privacy-problem-heres-how-to-fix-it\/\">Read More &raquo;<\/a><\/span><\/p>\n","protected":false},"author":1,"featured_media":18021,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_uag_custom_page_level_css":"","_umich_oidc_access":"","_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"_ef_editorial_meta_date_first-draft-date":"","_ef_editorial_meta_paragraph_assignment":"","_ef_editorial_meta_checkbox_needs-photo":"","_ef_editorial_meta_number_word-count":"","footnotes":""},"categories":[5],"tags":[598,121,130,154],"class_list":["post-18019","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-campus-news","tag-assistant","tag-privacy","tag-research","tag-voice"],"uagb_featured_image_src":{"full":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-e1583032525637.jpg",600,399,false],"thumbnail":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-200x133.jpg",200,133,true],"medium":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-300x200.jpg",300,200,true],"medium_large":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-768x511.jpg",665,442,true],"large":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-700x466.jpg",600,399,true],"1536x1536":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-1536x1022.jpg",1536,1022,true],"2048x2048":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-e1583032525637.jpg",600,399,false],"excerpt-thumbnail":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-200x140.jpg",200,140,true],"themonic-thumbnail":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-60x42.jpg",60,42,true],"ioslider-thumbnail":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-658x300.jpg",658,300,true],"post-thumbnail":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-665x442.jpg",665,442,true],"400x250-crop":["https:\/\/michigan.it.umich.edu\/news\/wp-content\/uploads\/2020\/02\/privacy-enhanced-content-e1583032525637.jpg",376,250,false]},"uagb_author_info":{"display_name":"News Staff","author_link":"https:\/\/michigan.it.umich.edu\/news\/author\/mitnewsadm\/"},"uagb_comment_info":0,"uagb_excerpt":"With devices listening everywhere you go, privacy concerns are endemic to advancing technology. For example, a virtual assistant that can learn to adapt to a user&#8217;s mood creates more useful, human-like interactions. But what if the audio powering these insights stores the user&#8217;s gender and demographic information on company servers \u2013 leaving the user open&hellip;","_links":{"self":[{"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/posts\/18019","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/comments?post=18019"}],"version-history":[{"count":5,"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/posts\/18019\/revisions"}],"predecessor-version":[{"id":18025,"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/posts\/18019\/revisions\/18025"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/media\/18021"}],"wp:attachment":[{"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/media?parent=18019"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/categories?post=18019"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/michigan.it.umich.edu\/news\/wp-json\/wp\/v2\/tags?post=18019"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}