{"id":"W4406330970","doi":"10.3390/s25020430","title":"In Shift and In Variance: Assessing the Robustness of HAR Deep Learning Models Against Variability","year":2025,"lang":"en","type":"article","venue":"Sensors","topic":"Context-Aware Activity Recognition Systems","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Concordia University","funders":"","keywords":"Robustness (evolution); Inertial measurement unit; Artificial intelligence; Computer science; Machine learning; Variance (accounting); Wearable computer; Statistics; Mathematics","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":[0.001645236,0.0001019922,0.0002364348,0.0001648641,0.00007242276,0.0001235318,0.0002929084,0.00007350459,0.000001937637],"category_scores_gemma":[0.0002900157,0.00008782448,0.00003060626,0.0007199889,0.0000660631,0.0006631765,0.000183389,0.000294858,8.743493e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006601793,"about_ca_system_score_gemma":0.00007992305,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001681012,"about_ca_topic_score_gemma":0.0001873929,"domain_scores_codex":[0.9982726,0.0007341488,0.0003218437,0.0003378007,0.0001500091,0.0001836221],"domain_scores_gemma":[0.9985387,0.0009387519,0.0001088357,0.0003284574,0.00006205511,0.0000232239],"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":[0.00001866876,0.0001722022,0.0423259,0.0002137359,0.00003026352,0.00003316459,0.007463031,0.8353912,0.0007479326,0.007977364,0.00001108405,0.1056155],"study_design_scores_gemma":[0.0002925681,0.00000457539,0.03401304,0.000146,0.000002437518,0.000003146357,0.0003343415,0.9614627,0.0001771073,0.003418779,0.00005562492,0.00008972762],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6629274,0.00004347788,0.3344473,0.0005875687,0.0001717446,0.0001528349,4.292e-7,0.0000251344,0.001644087],"genre_scores_gemma":[0.9982094,0.000008032594,0.001621043,0.00007539689,0.00001447153,0.00001112728,3.795385e-7,0.000004247186,0.00005588459],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.335282,"threshold_uncertainty_score":0.3581378,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02531544020292413,"score_gpt":0.2764825822121709,"score_spread":0.2511671420092468,"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."}}