{"id":"W4386465623","doi":"10.26434/chemrxiv-2023-zhlmr","title":"Global interpretability and geometry of graph convolu- tional neural networks for chemistry in terms of chemical moieties","year":2023,"lang":"en","type":"preprint","venue":"ChemRxiv","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of New Brunswick","funders":"Alliance de recherche numérique du Canada; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; Canada Foundation for Innovation; New Brunswick Innovation Foundation","keywords":"Embedding; Interpretability; Chemical space; Convolutional neural network; Artificial intelligence; Euclidean space; Computer science; Parametric statistics; Dimensionality reduction; Pattern recognition (psychology); Theoretical computer science; Algorithm; Mathematics; Chemistry; Combinatorics","routes":{"ca_aff":true,"ca_fund":true,"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"],"consensus_categories":[],"category_scores_codex":[0.001313232,0.0002909322,0.0007232412,0.0000622493,0.00002595449,0.000057148,0.0007564432,0.0003741057,0.0001650267],"category_scores_gemma":[0.0009188767,0.0002820457,0.0001545464,0.0002418822,0.0009928253,0.00007277664,0.001240963,0.0002791548,0.000001115675],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007551567,"about_ca_system_score_gemma":0.00006244519,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001514696,"about_ca_topic_score_gemma":0.00000594478,"domain_scores_codex":[0.9976792,0.00006880674,0.0007950511,0.0007738612,0.0003261571,0.0003568885],"domain_scores_gemma":[0.998444,0.0003108952,0.0005274963,0.0005243373,0.0001096005,0.00008369128],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0002198794,0.00009804712,0.1917217,0.00317061,0.00001625437,0.000001868828,0.0001403799,0.01950877,0.784727,0.000137934,0.0001328185,0.0001246995],"study_design_scores_gemma":[0.0009089307,0.00006926906,0.1057596,0.0007326551,0.0000600449,0.00001526282,0.00006020947,0.2016755,0.6604522,0.02953148,0.000009106734,0.0007257214],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9962562,0.0001044114,0.002145389,0.00009296113,0.0007638447,0.000338411,0.0001419272,0.00006798327,0.00008888556],"genre_scores_gemma":[0.9971284,0.000011375,0.002544109,0.00001448583,0.0001097952,0.00008453153,0.00006563659,0.00001942277,0.00002221474],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1821668,"threshold_uncertainty_score":0.9999632,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01566286879365314,"score_gpt":0.2796967816402281,"score_spread":0.264033912846575,"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."}}