{"id":"W3214261769","doi":"","title":"Self-Supervised Audio-Visual Representation Learning with Relaxed Cross-Modal Temporal Synchronicity","year":2021,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Queen's University","funders":"","keywords":"Synchronicity; Computer science; Modal; Modalities; Representation (politics); Artificial intelligence; Speech recognition; Asynchronous communication; Audio visual; Feature learning; Pattern recognition (psychology); Natural language processing; Machine learning; Psychology","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0003487972,0.0004311422,0.0004891046,0.0002026155,0.0005576168,0.0008389408,0.001281169,0.0003583657,0.00006599962],"category_scores_gemma":[0.00005664244,0.0004688854,0.0002188718,0.0009871599,0.0001629951,0.00130894,0.002052047,0.00120045,0.00003122708],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003831713,"about_ca_system_score_gemma":0.001011231,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002363201,"about_ca_topic_score_gemma":0.00004685413,"domain_scores_codex":[0.9967982,0.0003123949,0.0003010855,0.001820427,0.0002542593,0.0005135895],"domain_scores_gemma":[0.9978642,0.0001031034,0.0004722763,0.0009522488,0.0003986685,0.0002094702],"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":[0.0001618262,0.0004641288,0.2789598,0.0008366664,0.0005092137,0.003035298,0.004035263,0.6997111,0.00025664,0.008528475,0.000158422,0.00334316],"study_design_scores_gemma":[0.001217596,0.000123184,0.008405342,0.0002894583,0.00008718734,0.00002810694,0.000350247,0.9868432,0.0006447887,0.001021141,0.0002235023,0.0007662175],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5400579,0.00005926368,0.4575492,0.00006553971,0.0002379387,0.0001634197,0.00000108746,0.0004064627,0.001459142],"genre_scores_gemma":[0.9820849,0.00006346527,0.0161331,0.0001072859,0.0001347863,0.000001897741,0.00005006,0.00003184488,0.001392598],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.442027,"threshold_uncertainty_score":0.9997763,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04758006904470114,"score_gpt":0.2186203644114222,"score_spread":0.1710402953667211,"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."}}