{"id":"W3211965499","doi":"","title":"TriBERT: Human-centric Audio-visual Representation Learning","year":2021,"lang":"en","type":"article","venue":"Neural Information Processing Systems","topic":"Music and Audio Processing","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Toronto; University of British Columbia","funders":"","keywords":"Computer science; Audio visual; Representation (politics); Artificial intelligence; Human–computer interaction; Visualization; Computer vision; Speech recognition; Natural language processing; Multimedia","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":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0003344369,0.0001737326,0.0002320368,0.0001955745,0.0007502677,0.002282156,0.0003354666,0.00009250779,0.00001366731],"category_scores_gemma":[0.0001697795,0.0001658903,0.00006391752,0.001202525,0.00002784274,0.007106105,0.0001462442,0.0002671784,0.00008659053],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007989349,"about_ca_system_score_gemma":0.0001933342,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002020479,"about_ca_topic_score_gemma":6.835595e-7,"domain_scores_codex":[0.9979834,0.0001366617,0.0007051642,0.000273912,0.0005696188,0.0003312491],"domain_scores_gemma":[0.9985024,0.00004545676,0.0005815204,0.0002246784,0.0005491634,0.0000967304],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001461566,0.0001090709,0.004042342,0.001725114,0.00003615016,0.00005635558,0.01209149,0.03632863,0.006669976,0.008180653,0.007065225,0.9236804],"study_design_scores_gemma":[0.0008664066,0.00005531564,0.00126315,0.0002963322,0.00001425778,0.0003055386,0.001171082,0.9590924,0.006414427,0.0001543103,0.02993637,0.0004304236],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08954937,0.001237891,0.889638,0.00130129,0.001427408,0.0003096641,0.000001188249,0.0009852914,0.01554986],"genre_scores_gemma":[0.9966668,0.000007971647,0.001030782,0.0006277262,0.0001955217,0.00002524423,0.00005680852,0.000009010111,0.001380063],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.92325,"threshold_uncertainty_score":0.9987535,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02623141557770678,"score_gpt":0.2957809480702087,"score_spread":0.269549532492502,"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."}}