{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:DKQBQL22ZMM7CVWINQFX4UL4PE","short_pith_number":"pith:DKQBQL22","schema_version":"1.0","canonical_sha256":"1aa0182f5acb19f156c86c0b7e517c793a9c36433faac6afdaaa530e26df3e10","source":{"kind":"arxiv","id":"2206.08684","version":1},"attestation_state":"computed","paper":{"title":"Sparse Double Descent: Where Network Pruning Aggravates Overfitting","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Quanzhi Zhu, Zeke Xie, Zengchang Qin, Zheng He","submitted_at":"2022-06-17T11:02:15Z","abstract_excerpt":"People usually believe that network pruning not only reduces the computational cost of deep networks, but also prevents overfitting by decreasing model capacity. However, our work surprisingly discovers that network pruning sometimes even aggravates overfitting. We report an unexpected sparse double descent phenomenon that, as we increase model sparsity via network pruning, test performance first gets worse (due to overfitting), then gets better (due to relieved overfitting), and gets worse at last (due to forgetting useful information). While recent studies focused on the deep double descent "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2206.08684","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-06-17T11:02:15Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"75ba1e8a06676946519936e52cb199047117bc8d1e162931812490feab5b6272","abstract_canon_sha256":"0eafd97504ece806a3390592cf17bfd927175a62bf1f9dc6570bbee02a414b89"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:32:42.021299Z","signature_b64":"XuQIPxsxmjsNI4WTXjyXD1ui35g4oTNlg9qFhb/4DJ930Nn9rHArEeMX0uZr16CLJncCFQjCUXA8If85NWrWCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1aa0182f5acb19f156c86c0b7e517c793a9c36433faac6afdaaa530e26df3e10","last_reissued_at":"2026-07-05T04:32:42.020845Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:32:42.020845Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sparse Double Descent: Where Network Pruning Aggravates Overfitting","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Quanzhi Zhu, Zeke Xie, Zengchang Qin, Zheng He","submitted_at":"2022-06-17T11:02:15Z","abstract_excerpt":"People usually believe that network pruning not only reduces the computational cost of deep networks, but also prevents overfitting by decreasing model capacity. However, our work surprisingly discovers that network pruning sometimes even aggravates overfitting. We report an unexpected sparse double descent phenomenon that, as we increase model sparsity via network pruning, test performance first gets worse (due to overfitting), then gets better (due to relieved overfitting), and gets worse at last (due to forgetting useful information). While recent studies focused on the deep double descent "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2206.08684","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2206.08684/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2206.08684","created_at":"2026-07-05T04:32:42.020903+00:00"},{"alias_kind":"arxiv_version","alias_value":"2206.08684v1","created_at":"2026-07-05T04:32:42.020903+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2206.08684","created_at":"2026-07-05T04:32:42.020903+00:00"},{"alias_kind":"pith_short_12","alias_value":"DKQBQL22ZMM7","created_at":"2026-07-05T04:32:42.020903+00:00"},{"alias_kind":"pith_short_16","alias_value":"DKQBQL22ZMM7CVWI","created_at":"2026-07-05T04:32:42.020903+00:00"},{"alias_kind":"pith_short_8","alias_value":"DKQBQL22","created_at":"2026-07-05T04:32:42.020903+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2602.07618","citing_title":"Neural Networks With Dense Weights Are Not Universal Approximators","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2602.07618","citing_title":"Neural Networks With Dense Weights Are Not Universal Approximators","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DKQBQL22ZMM7CVWINQFX4UL4PE","json":"https://pith.science/pith/DKQBQL22ZMM7CVWINQFX4UL4PE.json","graph_json":"https://pith.science/api/pith-number/DKQBQL22ZMM7CVWINQFX4UL4PE/graph.json","events_json":"https://pith.science/api/pith-number/DKQBQL22ZMM7CVWINQFX4UL4PE/events.json","paper":"https://pith.science/paper/DKQBQL22"},"agent_actions":{"view_html":"https://pith.science/pith/DKQBQL22ZMM7CVWINQFX4UL4PE","download_json":"https://pith.science/pith/DKQBQL22ZMM7CVWINQFX4UL4PE.json","view_paper":"https://pith.science/paper/DKQBQL22","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2206.08684&json=true","fetch_graph":"https://pith.science/api/pith-number/DKQBQL22ZMM7CVWINQFX4UL4PE/graph.json","fetch_events":"https://pith.science/api/pith-number/DKQBQL22ZMM7CVWINQFX4UL4PE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DKQBQL22ZMM7CVWINQFX4UL4PE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DKQBQL22ZMM7CVWINQFX4UL4PE/action/storage_attestation","attest_author":"https://pith.science/pith/DKQBQL22ZMM7CVWINQFX4UL4PE/action/author_attestation","sign_citation":"https://pith.science/pith/DKQBQL22ZMM7CVWINQFX4UL4PE/action/citation_signature","submit_replication":"https://pith.science/pith/DKQBQL22ZMM7CVWINQFX4UL4PE/action/replication_record"}},"created_at":"2026-07-05T04:32:42.020903+00:00","updated_at":"2026-07-05T04:32:42.020903+00:00"}