{"id":"W4382319506","doi":"10.48550/arxiv.2306.13812","title":"Maintaining Plasticity in Deep Continual Learning","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Domain Adaptation and Few-Shot Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Alliance de recherche numérique du Canada; DeepMind; Alberta Machine Intelligence Institute; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"MNIST database; Backpropagation; Artificial intelligence; Computer science; Plasticity; Deep learning; Normalization (sociology); Regularization (linguistics); Machine learning; Artificial neural network; Task (project management); Engineering","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0005235354,0.0002951922,0.0003661814,0.0005620344,0.0001936214,0.0002086142,0.001324875,0.0002756365,0.0000452569],"category_scores_gemma":[0.0003558307,0.0003831841,0.0001497268,0.0008718282,0.00008719335,0.0003392581,0.002048794,0.001539757,0.0003595596],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002547293,"about_ca_system_score_gemma":0.0001652744,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001867257,"about_ca_topic_score_gemma":0.0002458855,"domain_scores_codex":[0.9976006,0.0003405916,0.0002801827,0.001110761,0.0001373734,0.0005305281],"domain_scores_gemma":[0.9985232,0.000494847,0.0002810226,0.0004324112,0.00009842568,0.0001700918],"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":[0.00001964095,0.00002575515,0.0123752,0.00002122155,0.00002778831,0.0007447062,0.001121059,0.8539057,0.00001554419,0.1306047,0.00002896125,0.001109823],"study_design_scores_gemma":[0.0006015491,0.00004074574,0.0121142,0.0001221209,0.00001381884,0.000003280594,0.0009775149,0.9717675,0.00001202578,0.01311185,0.0008179477,0.0004174031],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1757377,0.00001159349,0.8186213,0.00008792715,0.0004826813,0.0001601833,0.000001559408,0.0006034918,0.004293469],"genre_scores_gemma":[0.99352,0.00004872457,0.002719321,0.00007244986,0.00006839677,0.000001315412,0.00001230508,0.00002797752,0.003529497],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8177823,"threshold_uncertainty_score":0.999862,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09071131042013264,"score_gpt":0.2010276352872427,"score_spread":0.11031632486711,"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."}}