{"id":"W4318719663","doi":"10.48550/arxiv.2301.11990","title":"Alignment with human representations supports robust few-shot learning","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Office of Naval Research; Natural Sciences and Engineering Research Council of Canada","keywords":"Adversarial system; Computer science; Artificial intelligence; Domain (mathematical analysis); Shot (pellet); Machine learning; Mathematics","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"],"consensus_categories":[],"category_scores_codex":[0.0005285534,0.0004769974,0.0004521478,0.000474437,0.0007531497,0.0002986201,0.002154413,0.000285785,0.00009894174],"category_scores_gemma":[0.0000939067,0.0005377876,0.0002062798,0.001003457,0.0001764765,0.0005533065,0.003917991,0.001558857,0.0001571884],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003608557,"about_ca_system_score_gemma":0.0002491699,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006953676,"about_ca_topic_score_gemma":0.0001608798,"domain_scores_codex":[0.9964766,0.0003885178,0.0003227507,0.001909648,0.0002932275,0.0006092894],"domain_scores_gemma":[0.9971124,0.0002178393,0.0005363077,0.001707504,0.0001837504,0.0002421795],"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":[0.00001239562,0.00004649643,0.02086177,0.00004891401,0.0001494732,0.001293548,0.0005462297,0.9134998,0.00001759437,0.06307374,0.0003835085,0.00006649923],"study_design_scores_gemma":[0.001016284,0.0001783888,0.009340578,0.0002977256,0.0002422472,0.00002546077,0.0007466247,0.9665907,0.00007697974,0.01957283,0.0006828681,0.001229285],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06940732,0.00001115024,0.9227178,0.0003245662,0.0006375848,0.0004275274,0.000005114049,0.001215748,0.00525325],"genre_scores_gemma":[0.9786152,0.0000291643,0.00737157,0.00004287076,0.000145877,0.00000480682,0.0001000203,0.000068594,0.01362187],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9153461,"threshold_uncertainty_score":0.9997073,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1226970041717997,"score_gpt":0.2397631785504393,"score_spread":0.1170661743786396,"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."}}