{"id":"W4383047243","doi":"10.48550/arxiv.2306.17693","title":"Thompson sampling for improved exploration in GFlowNets","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Machine Learning and Algorithms","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Samsung; Genentech; Canadian Institute for Advanced Research","keywords":"Thompson sampling; Computer science; Flexibility (engineering); Sampling (signal processing); Mathematical optimization; Importance sampling; Convergence (economics); Inference; Machine learning; Generative grammar; Artificial intelligence; Posterior probability; Bayesian probability; Algorithm; Mathematics; Monte Carlo method; Statistics","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002652752,0.001070851,0.001492876,0.0009622758,0.0009197495,0.001192466,0.002345358,0.001833671,0.003738392],"category_scores_gemma":[0.01384845,0.0008019361,0.0008881777,0.000713131,0.001881418,0.002656049,0.002388315,0.002081093,0.0006109311],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001706813,"about_ca_system_score_gemma":0.001996102,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008486419,"about_ca_topic_score_gemma":0.01341207,"domain_scores_codex":[0.99907,0.00040741,0.00004435269,0.0001843482,0.0001615652,0.0001323125],"domain_scores_gemma":[0.9954875,0.003448491,0.0001970345,0.0003739643,0.0002958092,0.0001972353],"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":[0.0001968748,0.00007907166,0.001861638,0.00008471734,0.00004529942,0.0001159459,0.0002019021,0.8790732,0.001299698,0.05427982,0.002001503,0.0607604],"study_design_scores_gemma":[0.00001095409,0.00001570347,0.00004507277,0.000009463555,0.000003467818,0.000009722528,0.000007733673,0.9792224,0.0002521112,0.02011495,0.0003047484,0.000003730762],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03871051,0.0003210834,0.9566578,0.0003943435,0.00005210864,0.00007315548,0.0001085927,0.0008616638,0.002820834],"genre_scores_gemma":[0.6843596,0.0002391041,0.308946,0.0005076518,0.00006346849,0.0004143286,0.0005617658,0.0005295721,0.00437847],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008486419,"threshold_uncertainty_score":0.01687407,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1928492168762235,"score_gpt":0.2401838458746142,"score_spread":0.04733462899839069,"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."}}