{"id":"W3101466225","doi":"","title":"Recurrent neural networks learn robust representations by dynamically balancing compression and expansion","year":2019,"lang":"en","type":"preprint","venue":"","topic":"Neural dynamics and brain function","field":"Neuroscience","cited_by":22,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal","funders":"","keywords":"Curse of dimensionality; Computer science; Recurrent neural network; Stochastic gradient descent; Dimensionality reduction; Artificial intelligence; Generalization; Artificial neural network; Gradient descent; Task (project management); Chaotic; Dynamical systems theory; Machine learning; 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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.000158353,0.000359879,0.0003635558,0.0001048129,0.0002535807,0.0002953432,0.0003107092,0.0003044686,0.00007943541],"category_scores_gemma":[0.0001959634,0.0003026775,0.000118007,0.0001483839,0.00009424125,0.0001832117,0.001185872,0.001158563,0.00001277512],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007001731,"about_ca_system_score_gemma":0.00002685892,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000109552,"about_ca_topic_score_gemma":0.00001844806,"domain_scores_codex":[0.9972454,0.0002832077,0.0004417618,0.001257058,0.0003952441,0.0003773124],"domain_scores_gemma":[0.9985183,0.0004058285,0.0002740902,0.0005905298,0.00005758568,0.0001536451],"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":[0.0001523461,0.0001692742,0.002543301,0.0001662361,0.0000108655,0.00001597936,0.00007669239,0.8322123,0.1180679,0.0005182063,0.01162696,0.03443989],"study_design_scores_gemma":[0.0002849609,0.00009916707,0.001670302,0.0001339585,0.00001905227,0.000018023,0.00002360795,0.9963156,0.0006099896,0.0002076173,0.0002761498,0.0003415887],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8996268,0.0002226668,0.08857154,0.001707239,0.005385448,0.001278861,0.0000861373,0.0003563115,0.002764998],"genre_scores_gemma":[0.9957707,0.0005293021,0.0002732853,0.0007141484,0.0001300414,0.00003323023,0.0002590755,0.00004616343,0.002244084],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1641033,"threshold_uncertainty_score":0.9999425,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02942621118811315,"score_gpt":0.2733511511008031,"score_spread":0.24392493991269,"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."}}