{"id":"W3111605110","doi":"10.1609/aaai.v35i8.16859","title":"Provable Benefits of Overparameterization in Model Compression: From Double Descent to Pruning Neural Networks","year":2021,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Sparse and Compressive Sensing Techniques","field":"Engineering","cited_by":21,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"National Science Foundation","keywords":"Pruning; Computer science; Artificial neural network; Feature (linguistics); Deep neural networks; Parameterized complexity; Artificial intelligence; Machine learning; Algorithm","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":[0.007926587,0.001661798,0.001694736,0.001027383,0.0009935806,0.001872731,0.002221179,0.002366393,0.001787756],"category_scores_gemma":[0.08696115,0.001115583,0.0009989643,0.000836785,0.004456955,0.005172082,0.005704106,0.0057997,0.0003967864],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001392079,"about_ca_system_score_gemma":0.001263922,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001559004,"about_ca_topic_score_gemma":0.001731765,"domain_scores_codex":[0.9962252,0.001908156,0.0001819664,0.0005469191,0.0008565606,0.0002811602],"domain_scores_gemma":[0.9528137,0.03710758,0.002211252,0.005785385,0.001407758,0.000674326],"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":[0.0004592725,0.0002415634,0.007393306,0.0003541273,0.0001784885,0.0007538245,0.0006393765,0.6725382,0.008248529,0.2339535,0.005432114,0.06980781],"study_design_scores_gemma":[0.00002683787,0.00009179308,0.000533875,0.00006748719,0.00001802392,0.0001533735,0.00005082661,0.8755259,0.001567796,0.1211998,0.0007447865,0.00001959813],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08019719,0.00097116,0.9116947,0.002514915,0.00006408146,0.0000612599,0.0001223134,0.0006129516,0.003761472],"genre_scores_gemma":[0.8555787,0.001097162,0.1390403,0.001271488,0.0002351223,0.0002548815,0.0003315711,0.0004144089,0.001776397],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007926587,"threshold_uncertainty_score":0.0419203,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0892116449821859,"score_gpt":0.275641139043692,"score_spread":0.1864294940615061,"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."}}