{"id":8214,"date":"2026-02-19T05:35:00","date_gmt":"2026-02-18T20:35:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=8214"},"modified":"2026-02-14T21:38:41","modified_gmt":"2026-02-14T12:38:41","slug":"iv-co-scientist-multi-agent-llm-framework-for-causal-instrumental-variable-discovery","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=8214","title":{"rendered":"IV Co-Scientist: Multi-Agent LLM Framework for Causal Instrumental Variable Discovery"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>IV Co-Scientist: Multi-Agent LLM Framework for Causal Instrumental Variable Discovery\u00a0<\/strong>[61.2]<br>\u5185\u56e0\u6027\u5909\u6570\u3068\u7d50\u679c\u3068\u306e\u76f8\u540c\u6027\u306e\u5b58\u5728\u4e0b\u3067\u306f\u3001\u30a4\u30f3\u30b9\u30c8\u30a5\u30eb\u30e1\u30f3\u30bf\u30eb\u5909\u6570(IVs)\u3092\u7528\u3044\u3066\u5185\u56e0\u6027\u5909\u6570\u306e\u56e0\u679c\u52b9\u679c\u3092\u5206\u96e2\u3059\u308b\u3002 \u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb(LLM)\u304c\u3053\u306e\u8ab2\u984c\u306b\u6709\u52b9\u304b\u3069\u3046\u304b\u3092\u691c\u8a0e\u3059\u308b\u3002 \u672c\u7a3f\u3067\u306f,\u591a\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u30b7\u30b9\u30c6\u30e0\u3067\u3042\u308bIV Co-Scientist\u3092\u7d39\u4ecb\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2602.07943v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2602.07943v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Sun, 08 Feb 2026 12:28:29 GMT)<\/li>\n\n\n\n<li>\u300cin this paper, we investigate whether large language models can assist in the discovery of instrumental variables through a structured, multi-agent framework in which LLM-based agents propose, critique, and refine candidate instruments.\u300d\u3068instrumental variables\u3092\u767a\u898b\u3059\u308b\u305f\u3081\u306e\u30de\u30eb\u30c1\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u30b7\u30b9\u30c6\u30e0\u306e\u63d0\u6848\u3002\u300cOur empirical results on real-world data demonstrate that LLM-suggested instruments show meaningful consistency, providing a first step to- ward principled use of LLMs in variable discovery.\u300d\u3068\u4e00\u5b9a\u6709\u671b\u306a\u7d50\u679c\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":[42,623],"class_list":["post-8214","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-autonomous-agent","tag-623"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8214","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=8214"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8214\/revisions"}],"predecessor-version":[{"id":8215,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8214\/revisions\/8215"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=8214"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=8214"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=8214"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}