{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:T3UTY43KPFILQL7HZTP3QPMHYH","short_pith_number":"pith:T3UTY43K","schema_version":"1.0","canonical_sha256":"9ee93c736a7950b82fe7ccdfb83d87c1f6933849e51eb153c390da11c6af08fc","source":{"kind":"arxiv","id":"2208.05744","version":1},"attestation_state":"computed","paper":{"title":"On the Pros and Cons of Momentum Encoder in Self-Supervised Visual Representation Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Axi Niu, Chang D. Yoo, Chaoning Zhang, Kang Zhang, Trung Pham","submitted_at":"2022-08-11T10:44:49Z","abstract_excerpt":"Exponential Moving Average (EMA or momentum) is widely used in modern self-supervised learning (SSL) approaches, such as MoCo, for enhancing performance. We demonstrate that such momentum can also be plugged into momentum-free SSL frameworks, such as SimCLR, for a performance boost. Despite its wide use as a fundamental component in modern SSL frameworks, the benefit caused by momentum is not well understood. We find that its success can be at least partly attributed to the stability effect. In the first attempt, we analyze how EMA affects each part of the encoder and reveal that the portion n"},"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":"2208.05744","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2022-08-11T10:44:49Z","cross_cats_sorted":[],"title_canon_sha256":"ea81c80fdaea9c56c3c88937779b0ee70e6bab28e852c91327e49d46a014ab06","abstract_canon_sha256":"64a77825fc330425395535299dcb75a23c872d65a564bf3efd0fdb28404fc73b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:47:45.987624Z","signature_b64":"UOTIC5E28dNnyJMkPv2eLoANu1CWMfPTt2XL1d6gaHDN+/1F4tN6CTrfIhGUB6oUT6XKFX76YzV4m+72X1iACA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9ee93c736a7950b82fe7ccdfb83d87c1f6933849e51eb153c390da11c6af08fc","last_reissued_at":"2026-07-05T04:47:45.987269Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:47:45.987269Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Pros and Cons of Momentum Encoder in Self-Supervised Visual Representation Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Axi Niu, Chang D. Yoo, Chaoning Zhang, Kang Zhang, Trung Pham","submitted_at":"2022-08-11T10:44:49Z","abstract_excerpt":"Exponential Moving Average (EMA or momentum) is widely used in modern self-supervised learning (SSL) approaches, such as MoCo, for enhancing performance. We demonstrate that such momentum can also be plugged into momentum-free SSL frameworks, such as SimCLR, for a performance boost. Despite its wide use as a fundamental component in modern SSL frameworks, the benefit caused by momentum is not well understood. We find that its success can be at least partly attributed to the stability effect. In the first attempt, we analyze how EMA affects each part of the encoder and reveal that the portion n"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2208.05744","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/2208.05744/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":"2208.05744","created_at":"2026-07-05T04:47:45.987338+00:00"},{"alias_kind":"arxiv_version","alias_value":"2208.05744v1","created_at":"2026-07-05T04:47:45.987338+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2208.05744","created_at":"2026-07-05T04:47:45.987338+00:00"},{"alias_kind":"pith_short_12","alias_value":"T3UTY43KPFIL","created_at":"2026-07-05T04:47:45.987338+00:00"},{"alias_kind":"pith_short_16","alias_value":"T3UTY43KPFILQL7H","created_at":"2026-07-05T04:47:45.987338+00:00"},{"alias_kind":"pith_short_8","alias_value":"T3UTY43K","created_at":"2026-07-05T04:47:45.987338+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.09164","citing_title":"E-MD3C: Taming Masked Diffusion Transformers for Efficient Zero-Shot Object Customization","ref_index":2023,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/T3UTY43KPFILQL7HZTP3QPMHYH","json":"https://pith.science/pith/T3UTY43KPFILQL7HZTP3QPMHYH.json","graph_json":"https://pith.science/api/pith-number/T3UTY43KPFILQL7HZTP3QPMHYH/graph.json","events_json":"https://pith.science/api/pith-number/T3UTY43KPFILQL7HZTP3QPMHYH/events.json","paper":"https://pith.science/paper/T3UTY43K"},"agent_actions":{"view_html":"https://pith.science/pith/T3UTY43KPFILQL7HZTP3QPMHYH","download_json":"https://pith.science/pith/T3UTY43KPFILQL7HZTP3QPMHYH.json","view_paper":"https://pith.science/paper/T3UTY43K","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2208.05744&json=true","fetch_graph":"https://pith.science/api/pith-number/T3UTY43KPFILQL7HZTP3QPMHYH/graph.json","fetch_events":"https://pith.science/api/pith-number/T3UTY43KPFILQL7HZTP3QPMHYH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/T3UTY43KPFILQL7HZTP3QPMHYH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/T3UTY43KPFILQL7HZTP3QPMHYH/action/storage_attestation","attest_author":"https://pith.science/pith/T3UTY43KPFILQL7HZTP3QPMHYH/action/author_attestation","sign_citation":"https://pith.science/pith/T3UTY43KPFILQL7HZTP3QPMHYH/action/citation_signature","submit_replication":"https://pith.science/pith/T3UTY43KPFILQL7HZTP3QPMHYH/action/replication_record"}},"created_at":"2026-07-05T04:47:45.987338+00:00","updated_at":"2026-07-05T04:47:45.987338+00:00"}