{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:ZN4M5YMEWMSCQQOEF55TPX45DM","short_pith_number":"pith:ZN4M5YME","schema_version":"1.0","canonical_sha256":"cb78cee184b3242841c42f7b37df9d1b088ebbf4f90932b7ab091dcb8023b418","source":{"kind":"arxiv","id":"2012.00573","version":2},"attestation_state":"computed","paper":{"title":"Multi-level Knowledge Distillation via Knowledge Alignment and Correlation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Fei Ding, Feng Luo, Hongxin Hu, Venkat Krovi, Yin Yang","submitted_at":"2020-12-01T15:27:15Z","abstract_excerpt":"Knowledge distillation (KD) has become an important technique for model compression and knowledge transfer. In this work, we first perform a comprehensive analysis of the knowledge transferred by different KD methods. We demonstrate that traditional KD methods, which minimize the KL divergence of softmax outputs between networks, are related to the knowledge alignment of an individual sample only. Meanwhile, recent contrastive learning-based KD methods mainly transfer relational knowledge between different samples, namely, knowledge correlation. While it is important to transfer the full knowl"},"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":"2012.00573","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2020-12-01T15:27:15Z","cross_cats_sorted":[],"title_canon_sha256":"8ebcadf3f07669628547e0c5317f7c19e6661bc638589fe62f4f2e1d7204a9f0","abstract_canon_sha256":"198af612b09b9b36bdac6611447c3e2b34646bf0381cbfd4dfa221d5d4effda5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:46:11.127184Z","signature_b64":"dkVQ/ETEAMF4wKEbZEpVhrbY60RIEU3p0dVVe2vVvjLQVokDzLgIg9zuVbuT8pZoCg/5jR/DW+a3drmRbVMkAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cb78cee184b3242841c42f7b37df9d1b088ebbf4f90932b7ab091dcb8023b418","last_reissued_at":"2026-07-05T02:46:11.126761Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:46:11.126761Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Multi-level Knowledge Distillation via Knowledge Alignment and Correlation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Fei Ding, Feng Luo, Hongxin Hu, Venkat Krovi, Yin Yang","submitted_at":"2020-12-01T15:27:15Z","abstract_excerpt":"Knowledge distillation (KD) has become an important technique for model compression and knowledge transfer. In this work, we first perform a comprehensive analysis of the knowledge transferred by different KD methods. We demonstrate that traditional KD methods, which minimize the KL divergence of softmax outputs between networks, are related to the knowledge alignment of an individual sample only. Meanwhile, recent contrastive learning-based KD methods mainly transfer relational knowledge between different samples, namely, knowledge correlation. While it is important to transfer the full knowl"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2012.00573","kind":"arxiv","version":2},"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/2012.00573/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":"2012.00573","created_at":"2026-07-05T02:46:11.126810+00:00"},{"alias_kind":"arxiv_version","alias_value":"2012.00573v2","created_at":"2026-07-05T02:46:11.126810+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2012.00573","created_at":"2026-07-05T02:46:11.126810+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZN4M5YMEWMSC","created_at":"2026-07-05T02:46:11.126810+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZN4M5YMEWMSCQQOE","created_at":"2026-07-05T02:46:11.126810+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZN4M5YME","created_at":"2026-07-05T02:46:11.126810+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZN4M5YMEWMSCQQOEF55TPX45DM","json":"https://pith.science/pith/ZN4M5YMEWMSCQQOEF55TPX45DM.json","graph_json":"https://pith.science/api/pith-number/ZN4M5YMEWMSCQQOEF55TPX45DM/graph.json","events_json":"https://pith.science/api/pith-number/ZN4M5YMEWMSCQQOEF55TPX45DM/events.json","paper":"https://pith.science/paper/ZN4M5YME"},"agent_actions":{"view_html":"https://pith.science/pith/ZN4M5YMEWMSCQQOEF55TPX45DM","download_json":"https://pith.science/pith/ZN4M5YMEWMSCQQOEF55TPX45DM.json","view_paper":"https://pith.science/paper/ZN4M5YME","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2012.00573&json=true","fetch_graph":"https://pith.science/api/pith-number/ZN4M5YMEWMSCQQOEF55TPX45DM/graph.json","fetch_events":"https://pith.science/api/pith-number/ZN4M5YMEWMSCQQOEF55TPX45DM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZN4M5YMEWMSCQQOEF55TPX45DM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZN4M5YMEWMSCQQOEF55TPX45DM/action/storage_attestation","attest_author":"https://pith.science/pith/ZN4M5YMEWMSCQQOEF55TPX45DM/action/author_attestation","sign_citation":"https://pith.science/pith/ZN4M5YMEWMSCQQOEF55TPX45DM/action/citation_signature","submit_replication":"https://pith.science/pith/ZN4M5YMEWMSCQQOEF55TPX45DM/action/replication_record"}},"created_at":"2026-07-05T02:46:11.126810+00:00","updated_at":"2026-07-05T02:46:11.126810+00:00"}