{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000413209,0.0002874731,0.000343059,0.0002946713,0.0004321187,0.0003105101,0.001889648,0.0001329025,0.00001984694],"category_scores_gemma":[0.001184019,0.0002314859,0.0001517827,0.001216039,0.0002453596,0.000673204,0.0005557458,0.0002028647,0.000006685898],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001094401,"about_ca_system_score_gemma":0.0001610608,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002088391,"about_ca_topic_score_gemma":0.000004462877,"domain_scores_codex":[0.9978644,0.00002419049,0.0006509557,0.000659695,0.0004008055,0.0003999174],"domain_scores_gemma":[0.9971876,0.0002086291,0.0005127615,0.0003952374,0.001622864,0.00007288599],"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":[0.00009722644,0.0001385335,0.00006112189,0.00006125571,0.00002258438,7.394532e-8,0.0008738517,0.02264646,0.02620431,0.9039217,0.00009586475,0.04587701],"study_design_scores_gemma":[0.00008133779,0.00008586214,0.00004300907,0.0002233754,0.00001341909,7.845331e-7,0.0002529833,0.6294992,0.2681335,0.1014357,0.00006587076,0.0001649149],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.002940232,0.00001586871,0.9873288,0.002996156,0.0004909643,0.001066098,0.000007826864,0.0001663626,0.004987642],"genre_scores_gemma":[0.630717,0.00006286061,0.3674993,0.0003759821,0.00004407919,0.0001712893,0.000002500058,0.00001814317,0.00110894],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.802486,"threshold_uncertainty_score":0.943972,"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."}}