{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:SDBC6BVFQJRD5V2D3DNQVETT2Z","short_pith_number":"pith:SDBC6BVF","schema_version":"1.0","canonical_sha256":"90c22f06a582623ed743d8db0a9273d65c1799b5f854bef167adc20cac771af1","source":{"kind":"arxiv","id":"2502.14604","version":2},"attestation_state":"computed","paper":{"title":"Noisy Test-Time Adaptation in Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bo Han, Chentao Cao, Kun Zhang, Tongliang Liu, Yang Liu, Zhanke Zhou, Zhun Zhong","submitted_at":"2025-02-20T14:37:53Z","abstract_excerpt":"Test-time adaptation (TTA) aims to address distribution shifts between source and target data by relying solely on target data during testing. In open-world scenarios, models often encounter noisy samples, i.e., samples outside the in-distribution (ID) label space. Leveraging the zero-shot capability of pre-trained vision-language models (VLMs), this paper introduces Zero-Shot Noisy TTA (ZS-NTTA), focusing on adapting the model to target data with noisy samples during test-time in a zero-shot manner. We find existing TTA methods underperform under ZS-NTTA, often lagging behind even the frozen "},"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":"2502.14604","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-20T14:37:53Z","cross_cats_sorted":[],"title_canon_sha256":"285257a7c939cf06e70aa54929ef0f9bd7d5efeeab6b33358b38950f9a697924","abstract_canon_sha256":"13535deae0b78f700c59ddf78f1bee528cdc813118061cc656c69e5da182a2e6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:45:17.358836Z","signature_b64":"y9tb1uZxDF5iWl1exrHyzkBPnTE122QYyMNASmOTGt41sYrpzntJEMbNOkAwJClTCFLAirQwCni60HLBdudwBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"90c22f06a582623ed743d8db0a9273d65c1799b5f854bef167adc20cac771af1","last_reissued_at":"2026-07-05T10:45:17.358302Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:45:17.358302Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Noisy Test-Time Adaptation in Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bo Han, Chentao Cao, Kun Zhang, Tongliang Liu, Yang Liu, Zhanke Zhou, Zhun Zhong","submitted_at":"2025-02-20T14:37:53Z","abstract_excerpt":"Test-time adaptation (TTA) aims to address distribution shifts between source and target data by relying solely on target data during testing. In open-world scenarios, models often encounter noisy samples, i.e., samples outside the in-distribution (ID) label space. Leveraging the zero-shot capability of pre-trained vision-language models (VLMs), this paper introduces Zero-Shot Noisy TTA (ZS-NTTA), focusing on adapting the model to target data with noisy samples during test-time in a zero-shot manner. We find existing TTA methods underperform under ZS-NTTA, often lagging behind even the frozen "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.14604","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/2502.14604/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":"2502.14604","created_at":"2026-07-05T10:45:17.358375+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.14604v2","created_at":"2026-07-05T10:45:17.358375+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.14604","created_at":"2026-07-05T10:45:17.358375+00:00"},{"alias_kind":"pith_short_12","alias_value":"SDBC6BVFQJRD","created_at":"2026-07-05T10:45:17.358375+00:00"},{"alias_kind":"pith_short_16","alias_value":"SDBC6BVFQJRD5V2D","created_at":"2026-07-05T10:45:17.358375+00:00"},{"alias_kind":"pith_short_8","alias_value":"SDBC6BVF","created_at":"2026-07-05T10:45:17.358375+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25758","citing_title":"Dual Distribution Estimation for Zero-shot Noisy Test-Time Adaptation with VLMs","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SDBC6BVFQJRD5V2D3DNQVETT2Z","json":"https://pith.science/pith/SDBC6BVFQJRD5V2D3DNQVETT2Z.json","graph_json":"https://pith.science/api/pith-number/SDBC6BVFQJRD5V2D3DNQVETT2Z/graph.json","events_json":"https://pith.science/api/pith-number/SDBC6BVFQJRD5V2D3DNQVETT2Z/events.json","paper":"https://pith.science/paper/SDBC6BVF"},"agent_actions":{"view_html":"https://pith.science/pith/SDBC6BVFQJRD5V2D3DNQVETT2Z","download_json":"https://pith.science/pith/SDBC6BVFQJRD5V2D3DNQVETT2Z.json","view_paper":"https://pith.science/paper/SDBC6BVF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.14604&json=true","fetch_graph":"https://pith.science/api/pith-number/SDBC6BVFQJRD5V2D3DNQVETT2Z/graph.json","fetch_events":"https://pith.science/api/pith-number/SDBC6BVFQJRD5V2D3DNQVETT2Z/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SDBC6BVFQJRD5V2D3DNQVETT2Z/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SDBC6BVFQJRD5V2D3DNQVETT2Z/action/storage_attestation","attest_author":"https://pith.science/pith/SDBC6BVFQJRD5V2D3DNQVETT2Z/action/author_attestation","sign_citation":"https://pith.science/pith/SDBC6BVFQJRD5V2D3DNQVETT2Z/action/citation_signature","submit_replication":"https://pith.science/pith/SDBC6BVFQJRD5V2D3DNQVETT2Z/action/replication_record"}},"created_at":"2026-07-05T10:45:17.358375+00:00","updated_at":"2026-07-05T10:45:17.358375+00:00"}