{"id":"W4392904080","doi":"10.1109/icassp48485.2024.10447379","title":"Engineering the Neural Collapse Geometry of Supervised-Contrastive Loss","year":2024,"lang":"en","type":"article","venue":"","topic":"Domain Adaptation and Few-Shot Learning","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"","keywords":"Embedding; Limiting; Computer science; Artificial intelligence; Entropy (arrow of time); Classifier (UML); Benchmark (surveying); Feature (linguistics); Artificial neural network; Geometry; Feature vector; Cross entropy; Machine learning; Pattern recognition (psychology); Algorithm; Mathematics; Engineering; Mechanical engineering","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002354157,0.00007136421,0.00007957948,0.00009419674,0.00004187737,0.0001599805,0.0003659589,0.00002184221,0.00009645139],"category_scores_gemma":[0.00006446913,0.00004703021,0.00005311746,0.0006213067,0.00002681756,0.00025928,0.00009137551,0.0001228807,0.00005611175],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001347069,"about_ca_system_score_gemma":0.00003331103,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001218575,"about_ca_topic_score_gemma":0.000001514701,"domain_scores_codex":[0.999367,0.00002419723,0.0001327444,0.0001611373,0.0001704688,0.0001444623],"domain_scores_gemma":[0.9993969,0.0003278901,0.00001523135,0.0001805802,0.00003865543,0.00004080713],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001039975,0.00004204843,0.0008747934,0.0001116124,0.0001128051,0.0001160442,0.006548243,0.0432969,0.006715333,0.804374,0.002027298,0.1357705],"study_design_scores_gemma":[0.00008913226,0.00002161766,0.003161047,0.00001628208,0.000003505029,0.00001692679,0.0001427429,0.9893939,0.0008456194,0.0001388299,0.006099168,0.00007119906],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07603876,0.0005409294,0.9162921,0.001759941,0.0006838975,0.0001067244,0.000001541268,0.0002740105,0.004302076],"genre_scores_gemma":[0.9928442,0.00000700931,0.005979215,0.0001953852,0.00004225579,0.000004269052,5.104156e-7,0.000005979004,0.0009211291],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.946097,"threshold_uncertainty_score":0.1917836,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01037520371146061,"score_gpt":0.2223468966375305,"score_spread":0.2119716929260698,"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."}}