{"id":"W4409364742","doi":"10.1609/aaai.v39i13.33576","title":"Forward KL Regularized Preference Optimization for Aligning Diffusion Policies","year":2025,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Advanced Multi-Objective Optimization Algorithms","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"National Natural Science Foundation of China","keywords":"Preference; Diffusion; Computer science; Mathematical optimization; Mathematics; Statistics; Physics; Thermodynamics","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":[0.002346492,0.001550689,0.001490708,0.0008744851,0.000544067,0.001239654,0.001332155,0.001729635,0.003008305],"category_scores_gemma":[0.008027262,0.0007929636,0.0009805023,0.001023265,0.001181533,0.002846031,0.001703915,0.00282319,0.0008177913],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001337611,"about_ca_system_score_gemma":0.002087411,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003981606,"about_ca_topic_score_gemma":0.004583834,"domain_scores_codex":[0.9985878,0.0005122615,0.00009821427,0.0003555986,0.0002901004,0.0001559826],"domain_scores_gemma":[0.9974819,0.001467882,0.0002774352,0.000272536,0.000356719,0.0001436024],"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":[0.0001084231,0.0001046128,0.0008106774,0.00008765625,0.00003980309,0.00006001147,0.0000670688,0.9050245,0.002618057,0.01023036,0.001959916,0.07888887],"study_design_scores_gemma":[0.000009445017,0.00002260672,0.0000444389,0.000004107048,0.000002830084,0.0000103477,0.000006420483,0.9942008,0.0006194442,0.004865814,0.0002089581,0.00000477168],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01895857,0.0001758643,0.9782668,0.0002398726,0.00003579905,0.00005663696,0.00008133661,0.0008673052,0.001317784],"genre_scores_gemma":[0.6584073,0.0002309993,0.3344951,0.000538095,0.00006723336,0.0003169397,0.0005901175,0.0005237301,0.004830477],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003981606,"threshold_uncertainty_score":0.01240957,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0602084729461822,"score_gpt":0.310791092154396,"score_spread":0.2505826192082138,"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."}}