{"id":"W3186092959","doi":"10.48550/arxiv.2107.10211","title":"Differentiable Annealed Importance Sampling and the Perils of Gradient Noise","year":2021,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Markov Chains and Monte Carlo Methods","field":"Mathematics","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Differentiable function; Computation; Mathematical optimization; Convergence (economics); Sublinear function; Noise (video); Mathematics; Sampling (signal processing); Computer science; Rate of convergence; Applied mathematics; Algorithm; Artificial intelligence","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005465176,0.0009191096,0.001268276,0.0008374075,0.0007475049,0.001774929,0.002149997,0.001844295,0.002810264],"category_scores_gemma":[0.03471281,0.0009169094,0.0007828716,0.0007845265,0.003792054,0.003048985,0.00226175,0.003462124,0.000517219],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002004054,"about_ca_system_score_gemma":0.001632708,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003215087,"about_ca_topic_score_gemma":0.003857067,"domain_scores_codex":[0.9976463,0.001269775,0.00008709195,0.000388243,0.0004581336,0.0001505054],"domain_scores_gemma":[0.9843653,0.01217132,0.0009297229,0.00136454,0.0007825883,0.0003864265],"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":[0.0001347221,0.00005531234,0.001740643,0.0001880845,0.0000802577,0.0001596943,0.000199788,0.341794,0.00207208,0.6275073,0.002034667,0.02403357],"study_design_scores_gemma":[0.00001747092,0.0000271949,0.0002233151,0.00002615939,0.00001068074,0.00003141978,0.00001007206,0.826617,0.0007750099,0.1711064,0.001140718,0.00001465034],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01269095,0.0004553068,0.9838066,0.0005889495,0.00006573155,0.000031971,0.00004831502,0.0002389994,0.002073133],"genre_scores_gemma":[0.6503001,0.001013693,0.3379439,0.0008209077,0.0002540989,0.000332735,0.0002690662,0.0004064336,0.008659097],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005465176,"threshold_uncertainty_score":0.02890301,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1663564451730863,"score_gpt":0.2479423587077539,"score_spread":0.0815859135346676,"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."}}