{"id":7134,"date":"2025-07-23T06:20:00","date_gmt":"2025-07-22T21:20:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7134"},"modified":"2025-07-20T15:24:22","modified_gmt":"2025-07-20T06:24:22","slug":"conformal-prediction-for-privacy-preserving-machine-learning","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7134","title":{"rendered":"Conformal Prediction for Privacy-Preserving Machine Learning"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Conformal Prediction for Privacy-Preserving Machine Learning&nbsp;<\/strong>[83.9]<br>AES\u3067\u6697\u53f7\u5316\u3055\u308c\u305fMNIST\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306e\u5909\u7a2e\u3092\u7528\u3044\u3066\u3001\u6697\u53f7\u5316\u3055\u308c\u305f\u30c9\u30e1\u30a4\u30f3\u306b\u76f4\u63a5\u9069\u7528\u3057\u3066\u3082\u3001\u30b3\u30f3\u30d5\u30a9\u30fc\u30de\u30eb\u4e88\u6e2c\u6cd5\u304c\u6709\u52b9\u3067\u3042\u308b\u3053\u3068\u3092\u793a\u3059\u3002 \u6211\u3005\u306e\u7814\u7a76\u306f\u3001\u5b89\u5168\u3067\u30d7\u30e9\u30a4\u30d0\u30b7\u30fc\u306b\u914d\u616e\u3057\u305f\u5b66\u7fd2\u30b7\u30b9\u30c6\u30e0\u306b\u304a\u3051\u308b\u539f\u5247\u7684\u4e0d\u78ba\u5b9f\u6027\u5b9a\u91cf\u5316\u306e\u57fa\u790e\u3092\u5b9a\u3081\u3066\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2507.09678v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2507.09678v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Sun, 13 Jul 2025 15:29:14 GMT)<\/li>\n\n\n\n<li>\u300cWe then assess the same model architecture under encryption. When trained on MNIST images encrypted with a fixed key and initialization vector (AES encryption; see Section 3), the model attains an average training accuracy of 39.48% and a test accuracy of 36.88%.\u300d\u3063\u3066\u672c\u5f53\u306a\u3093\u3060\u308d\u3046\u304b\u2026\u300cIn contrast, training the same model on the MNIST dataset with randomized encryption per sample (a unique key per image) results in a test accuracy of 9.56%, indistinguishable from random guessing.\u300d\u3068\u8a18\u8f09\u3055\u308c\u3066\u3044\u308b\u3068\u3044\u3046\u3053\u3068\u306fleak\u3068\u304b\u3067\u306f\u306a\u3055\u305d\u3046\u3060\u304c\u3002\u3002\u3002\u30ad\u30fc\u3068IV\u304c\u56fa\u5b9a\u3068\u306f\u3044\u3048\u3001\u7d50\u69cb\u9a5a\u304d\u304c\u3042\u308b\u3002<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[688],"class_list":["post-7134","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-688"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7134","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=7134"}],"version-history":[{"count":2,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7134\/revisions"}],"predecessor-version":[{"id":7136,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7134\/revisions\/7136"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7134"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7134"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7134"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}