{"id":"W4414510082","doi":"10.21203/rs.3.rs-7540516/v1","title":"BiRLNN: Bidirectional Reinforcement-Learning Neural Network for Constrained Molecular Design","year":2025,"lang":"en","type":"preprint","venue":"Research Square","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"Defence Research and Development Canada; National Research Council Canada","funders":"","keywords":"Reinforcement learning; Process (computing); Constraint (computer-aided design); Space (punctuation); Artificial neural network; Chemical space; Function (biology)","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":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.01045446,0.0004627265,0.0006206676,0.0005005566,0.001177049,0.00097502,0.001519128,0.0004298234,0.001831798],"category_scores_gemma":[0.004364859,0.0004515931,0.0002472874,0.0007168599,0.0005947232,0.0001510819,0.002063671,0.001655485,0.0001463665],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003315202,"about_ca_system_score_gemma":0.001375407,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003105842,"about_ca_topic_score_gemma":0.000008430249,"domain_scores_codex":[0.9917464,0.002715226,0.0007257464,0.001380606,0.001802016,0.001629988],"domain_scores_gemma":[0.9952694,0.002184128,0.0003234341,0.0008408461,0.001118539,0.0002636862],"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":[0.0002373216,0.00002699056,0.0001909114,0.000824651,0.0000215015,0.00002006988,0.0001198797,0.9064622,0.08439004,0.002760686,0.004445187,0.000500589],"study_design_scores_gemma":[0.001004703,0.0007893399,0.0003169168,0.001610495,0.00004505565,0.00001617391,0.0001311719,0.9255101,0.04560455,0.01199747,0.01203384,0.0009401942],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.094573,0.001183026,0.8758126,0.002556279,0.006999846,0.008851792,0.0003075363,0.001396137,0.008319819],"genre_scores_gemma":[0.8322976,0.0001034819,0.1526552,0.0002541426,0.001696955,0.002780305,0.000454216,0.0001200015,0.009638073],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.7377246,"threshold_uncertainty_score":0.9997936,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06354384895276662,"score_gpt":0.3933398122354619,"score_spread":0.3297959632826953,"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."}}