{"id":"W4409035697","doi":"10.54364/aaiml.2025.51205","title":"Improving CNN Robustness to Color Shifts via Color Balancing and Spatial Dropout","year":2025,"lang":"en","type":"article","venue":"Advances in Artificial Intelligence and Machine Learning","topic":"Color Science and Applications","field":"Physics and Astronomy","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Canadian Institute for Advanced Research","keywords":"Robustness (evolution); Computer science; Artificial intelligence; Dropout (neural networks); Computer vision; Pattern recognition (psychology); Machine learning","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002530769,0.0001392007,0.000185136,0.0001296695,0.0003510146,0.0001021373,0.0001385987,0.00003026404,0.00004782013],"category_scores_gemma":[0.00004789922,0.0001355743,0.00002482924,0.0004320396,0.00009972119,0.0002652731,0.0001568421,0.0002496876,0.000009761103],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002112306,"about_ca_system_score_gemma":0.00003586859,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001836553,"about_ca_topic_score_gemma":0.002105443,"domain_scores_codex":[0.998936,0.00003485419,0.0002835864,0.0003818189,0.00008687418,0.0002768685],"domain_scores_gemma":[0.9995473,0.0001604962,0.00007506896,0.0001062748,0.00003716032,0.00007367656],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00002845524,0.00003835261,0.03675316,0.00001450779,0.000003915006,0.000001032881,0.0003964343,0.04503878,0.003377068,0.02183354,0.000001347979,0.8925134],"study_design_scores_gemma":[0.00007796669,0.0001298756,0.001796859,0.00008626183,0.0000172477,6.565556e-7,0.002018761,0.9622707,0.01063711,0.01669118,0.005978157,0.0002951612],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3888752,0.0002546987,0.6091225,0.0005088766,0.0001181676,0.000231407,0.000002622524,0.00002383997,0.0008626583],"genre_scores_gemma":[0.9977978,0.0000420529,0.001700946,0.0001051731,0.00008524267,0.000076632,0.000005846888,0.000006971857,0.0001793694],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.917232,"threshold_uncertainty_score":0.5528561,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0112371400466142,"score_gpt":0.2975237435100667,"score_spread":0.2862866034634525,"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."}}