{"id":"W4280643089","doi":"10.1609/aaai.v36i6.20558","title":"Feedback Gradient Descent: Efficient and Stable Optimization with Orthogonality for DNNs","year":2022,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Human Pose and Action Recognition","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"Defense Acquisition Program Administration; Agency for Defense Development","keywords":"Stiefel manifold; Orthogonality; Tangent bundle; Stability (learning theory); Gradient descent; Manifold (fluid mechanics); Computer science; Discretization; Mathematical optimization; Tangent vector; Mathematics; Applied mathematics; Tangent; Control theory (sociology); Algorithm; Tangent space; Artificial neural network; Artificial intelligence; Mathematical analysis; Machine learning; Geometry","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001284116,0.001279879,0.001001459,0.0005844526,0.000439467,0.0007326464,0.001319714,0.001197156,0.001696142],"category_scores_gemma":[0.004086895,0.0006235021,0.0005710334,0.0005652601,0.0009286702,0.001271881,0.001642804,0.001502459,0.0005852744],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001055812,"about_ca_system_score_gemma":0.001715915,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005159281,"about_ca_topic_score_gemma":0.006451944,"domain_scores_codex":[0.9994772,0.0001581364,0.00003440404,0.0000924604,0.0001846372,0.00005311839],"domain_scores_gemma":[0.9993193,0.0003013291,0.00007227154,0.00008359695,0.0001672849,0.00005622505],"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":[0.0001708979,0.0001033718,0.0008708816,0.0001448915,0.00007722544,0.0001054596,0.00009994801,0.7662828,0.01218425,0.03421343,0.004160533,0.1815863],"study_design_scores_gemma":[0.0000054276,0.00001081012,0.00002375422,0.000003135887,0.000001620406,0.000005752645,0.000002126683,0.9961969,0.0005969667,0.002887137,0.0002640878,0.000002276186],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.008492569,0.0001785537,0.989575,0.00009988066,0.00003845623,0.000036612,0.00003619871,0.0006481159,0.0008946703],"genre_scores_gemma":[0.3791323,0.0003207868,0.6156759,0.0002663052,0.00007321379,0.0002833035,0.000327801,0.00047228,0.003448053],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005159281,"threshold_uncertainty_score":0.0102585,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06031393677450669,"score_gpt":0.2617351034577038,"score_spread":0.2014211666831971,"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."}}