{"id":8143,"date":"2026-01-26T05:37:00","date_gmt":"2026-01-25T20:37:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=8143"},"modified":"2026-01-25T09:39:19","modified_gmt":"2026-01-25T00:39:19","slug":"agentic-confidence-calibration","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=8143","title":{"rendered":"Agentic Confidence Calibration"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Agentic Confidence Calibration&nbsp;<\/strong>[67.5]<br>Holistic Trajectory (HTC)\u306fAI\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u306e\u65b0\u3057\u3044\u8a3a\u65ad\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3067\u3042\u308b\u3002 HTC\u306f\u30ad\u30e3\u30ea\u30d6\u30ec\u30fc\u30b7\u30e7\u30f3\u3068\u5dee\u5225\u306e\u4e21\u65b9\u306b\u304a\u3044\u3066\u3001\u5f37\u529b\u306a\u30d9\u30fc\u30b9\u30e9\u30a4\u30f3\u3092\u4e00\u8cab\u3057\u3066\u8d85\u3048\u3066\u3044\u308b\u3002 HTC\u306f\u3001\u969c\u5bb3\u306e\u80cc\u5f8c\u306b\u3042\u308b\u30b7\u30b0\u30ca\u30eb\u3092\u660e\u3089\u304b\u306b\u3059\u308b\u3053\u3068\u306b\u3088\u3063\u3066\u3001\u89e3\u91c8\u53ef\u80fd\u6027\u3092\u63d0\u4f9b\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2601.15778v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2601.15778v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Thu, 22 Jan 2026 09:08:25 GMT)<\/li>\n\n\n\n<li>\u300cOur work addresses compounding uncertainty, heterogeneous signals, and data scarcity, yielding three key takeaways: (1) calibration relies on a hierarchy of diagnostic signals; (2) HTC features capture a transferable \u201cuncertainty patterns\u201d enabling strong cross-task generalization while exposing limits under distribution shift; and (3) a pretrained General Agent Calibrator (GAC) achieves the best ECE (zero-shot) on unseen tasks like GAIA, providing a plug-and-play foundation.\u300d\u3068\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u306e\u4fe1\u983c\u5ea6\u3092\u7dcf\u5408\u7684\u306b\u8a55\u4fa1\u3057\u3066\u3044\u304f\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af<\/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":[223,648],"class_list":["post-8143","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-648"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8143","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=8143"}],"version-history":[{"count":2,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8143\/revisions"}],"predecessor-version":[{"id":8155,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8143\/revisions\/8155"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=8143"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=8143"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=8143"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}