{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FIALIQQST7G6HMQEOCIAUQME5H","short_pith_number":"pith:FIALIQQS","schema_version":"1.0","canonical_sha256":"2a00b442129fcde3b20470900a4184e9edbc76ef6f9ad4fba84e1d710116515c","source":{"kind":"arxiv","id":"2606.01908","version":1},"attestation_state":"computed","paper":{"title":"Private and Stable Test-Time Adaptation with Differential Privacy","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Evan Shelhamer, Mathias Lecuyer, Qiaoyue Tang, Zefeng Li","submitted_at":"2026-06-01T08:45:59Z","abstract_excerpt":"Test-time adaptation (TTA) can reduce error on new and different data by updating the model on these inputs during inference. However, these updates raise the issue of privacy w.r.t. the testing data, because the model parameters now depend on all past inputs. To control this privacy risk, we cast multiple popular TTA methods (Tent, EATA, SAR, DeYO, and COME) into differential privacy (DP) forms that apply per-sample gradient clipping and Gaussian noise for all updates. On ImageNet-C, our DP-TTA methods provide adequate privacy at small cost to accuracy, and in the low-privacy regime the clipp"},"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":"2606.01908","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-01T08:45:59Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"fdd4e5454657c3f14b1d5228469507d4cd8824cedf3a8ad5837d9167bb4d259a","abstract_canon_sha256":"7f75af9430bfb966492cb14ab2fcb53bb3d8f2c15d34909264383f6b8438bf56"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:05:00.137610Z","signature_b64":"DdlgmX2+fkfZPlZZLWE1Q1MnvWZR/KPp4kL9kJM+uTLyR4b9zVeUd3MDzC/Jk97AtCnZsxr7ODYXFWfdLjDjAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2a00b442129fcde3b20470900a4184e9edbc76ef6f9ad4fba84e1d710116515c","last_reissued_at":"2026-06-02T02:05:00.137176Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:05:00.137176Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Private and Stable Test-Time Adaptation with Differential Privacy","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Evan Shelhamer, Mathias Lecuyer, Qiaoyue Tang, Zefeng Li","submitted_at":"2026-06-01T08:45:59Z","abstract_excerpt":"Test-time adaptation (TTA) can reduce error on new and different data by updating the model on these inputs during inference. However, these updates raise the issue of privacy w.r.t. the testing data, because the model parameters now depend on all past inputs. To control this privacy risk, we cast multiple popular TTA methods (Tent, EATA, SAR, DeYO, and COME) into differential privacy (DP) forms that apply per-sample gradient clipping and Gaussian noise for all updates. On ImageNet-C, our DP-TTA methods provide adequate privacy at small cost to accuracy, and in the low-privacy regime the clipp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.01908","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/2606.01908/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":"2606.01908","created_at":"2026-06-02T02:05:00.137268+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.01908v1","created_at":"2026-06-02T02:05:00.137268+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.01908","created_at":"2026-06-02T02:05:00.137268+00:00"},{"alias_kind":"pith_short_12","alias_value":"FIALIQQST7G6","created_at":"2026-06-02T02:05:00.137268+00:00"},{"alias_kind":"pith_short_16","alias_value":"FIALIQQST7G6HMQE","created_at":"2026-06-02T02:05:00.137268+00:00"},{"alias_kind":"pith_short_8","alias_value":"FIALIQQS","created_at":"2026-06-02T02:05:00.137268+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/FIALIQQST7G6HMQEOCIAUQME5H","json":"https://pith.science/pith/FIALIQQST7G6HMQEOCIAUQME5H.json","graph_json":"https://pith.science/api/pith-number/FIALIQQST7G6HMQEOCIAUQME5H/graph.json","events_json":"https://pith.science/api/pith-number/FIALIQQST7G6HMQEOCIAUQME5H/events.json","paper":"https://pith.science/paper/FIALIQQS"},"agent_actions":{"view_html":"https://pith.science/pith/FIALIQQST7G6HMQEOCIAUQME5H","download_json":"https://pith.science/pith/FIALIQQST7G6HMQEOCIAUQME5H.json","view_paper":"https://pith.science/paper/FIALIQQS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.01908&json=true","fetch_graph":"https://pith.science/api/pith-number/FIALIQQST7G6HMQEOCIAUQME5H/graph.json","fetch_events":"https://pith.science/api/pith-number/FIALIQQST7G6HMQEOCIAUQME5H/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FIALIQQST7G6HMQEOCIAUQME5H/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FIALIQQST7G6HMQEOCIAUQME5H/action/storage_attestation","attest_author":"https://pith.science/pith/FIALIQQST7G6HMQEOCIAUQME5H/action/author_attestation","sign_citation":"https://pith.science/pith/FIALIQQST7G6HMQEOCIAUQME5H/action/citation_signature","submit_replication":"https://pith.science/pith/FIALIQQST7G6HMQEOCIAUQME5H/action/replication_record"}},"created_at":"2026-06-02T02:05:00.137268+00:00","updated_at":"2026-06-02T02:05:00.137268+00:00"}