{"id":"W4383604828","doi":"10.48550/arxiv.2307.02849","title":"NatLogAttack: A Framework for Attacking Natural Language Inference Models with Natural Logic","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Adversarial system; Computer science; Inference; Artificial intelligence; Spurious relationship; Natural language; Formalism (music); Natural language understanding; Non-monotonic logic; Natural (archaeology); Cognitive science; Machine learning; Psychology","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.0003884717,0.0005958405,0.0005884949,0.0004113362,0.0003824823,0.0003204078,0.003377948,0.00049765,0.000007610836],"category_scores_gemma":[0.0004660155,0.0005847199,0.0002862613,0.001024387,0.0001778954,0.0008906195,0.003610573,0.002477117,0.0000394724],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003309869,"about_ca_system_score_gemma":0.0003128897,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001754711,"about_ca_topic_score_gemma":0.0001246883,"domain_scores_codex":[0.9967232,0.0001886982,0.0002682347,0.001816526,0.0002338693,0.0007695228],"domain_scores_gemma":[0.9963412,0.001190569,0.0004761199,0.001527665,0.0003089287,0.0001555255],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00004964104,0.00001599143,0.0004018546,0.00009093124,0.00008210644,0.000285743,0.0005057937,0.5820404,0.000004129375,0.4161777,0.00002706927,0.0003186249],"study_design_scores_gemma":[0.0004602389,0.00005291495,0.0002263967,0.0003775475,0.00006087322,0.000006056604,0.0002307427,0.7957372,0.00001525146,0.2021936,0.00001855877,0.0006206566],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04075784,0.0002049559,0.9549684,0.0002967588,0.001834798,0.000619785,0.00001657601,0.0010843,0.0002165825],"genre_scores_gemma":[0.8686049,0.00003161481,0.1296227,0.0001771907,0.0002034321,0.000006194814,0.00004731508,0.00005789981,0.001248785],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8278471,"threshold_uncertainty_score":0.9998242,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1167031844234035,"score_gpt":0.2654593571134707,"score_spread":0.1487561726900672,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}