{"id":"W3123796542","doi":"10.3389/fnhum.2021.653659","title":"BENDR: Using Transformers and a Contrastive Self-Supervised Learning Task to Learn From Massive Amounts of EEG Data","year":2021,"lang":"en","type":"preprint","venue":"Frontiers in Human Neuroscience","topic":"EEG and Brain-Computer Interfaces","field":"Neuroscience","cited_by":11,"is_retracted":false,"has_abstract":true,"ca_institutions":"St. Michael's Hospital; Vector Institute; University of Toronto","funders":"Electronics and Telecommunications Research Institute; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Computer science; Electroencephalography; Artificial intelligence; Transformer; Raw data; Leverage (statistics); Variety (cybernetics); Brain–computer interface; Machine learning; Task (project management); Inference; Labeled data; Speech recognition; Pattern recognition (psychology)","routes":{"ca_aff":true,"ca_fund":true,"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":[0.0004336793,0.0005199548,0.0008827621,0.0005030258,0.0003522534,0.0004823729,0.00230153,0.0002217451,0.00001398747],"category_scores_gemma":[0.0007359274,0.0005541741,0.0001002569,0.0006296533,0.000688507,0.0006895888,0.002507221,0.001339873,3.941956e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001328927,"about_ca_system_score_gemma":0.0002641079,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003935619,"about_ca_topic_score_gemma":0.00005710491,"domain_scores_codex":[0.994592,0.0005886734,0.0007256378,0.002646043,0.0007551931,0.0006924714],"domain_scores_gemma":[0.9981702,0.0002391856,0.0003798836,0.0008830287,0.00007879949,0.0002488932],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00007068481,0.000205176,0.01030561,0.0001864997,0.00001809378,0.0002195074,0.009512444,0.03901439,0.9358644,0.00001946936,0.0002372055,0.00434649],"study_design_scores_gemma":[0.001233373,0.0003800576,0.009637265,0.001075583,0.0001027063,0.00002208425,0.002704697,0.8904428,0.09227146,0.0003181182,0.0006592338,0.001152577],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8896158,0.0002108666,0.1062698,0.0001171646,0.002432032,0.0007180021,0.0004138874,0.00008230382,0.0001400918],"genre_scores_gemma":[0.9889153,0.0001414468,0.01019139,0.0005303804,0.0000601209,0.00001606231,0.00003572884,0.00004818973,0.00006141884],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8514284,"threshold_uncertainty_score":0.9996909,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05303828779676278,"score_gpt":0.2995327502607941,"score_spread":0.2464944624640313,"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."}}