{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:7E4Y55ZGVSESCL6VW7DQMX34FK","short_pith_number":"pith:7E4Y55ZG","schema_version":"1.0","canonical_sha256":"f9398ef726ac89212fd5b7c7065f7c2ab7ed59581c4cf7cdd418978014043b92","source":{"kind":"arxiv","id":"2407.14879","version":1},"attestation_state":"computed","paper":{"title":"Thompson Sampling Itself is Differentially Private","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DS","stat.ML"],"primary_cat":"cs.LG","authors_text":"Marco Avella Medina, Rachel Cummings, Tingting Ou","submitted_at":"2024-07-20T14:01:03Z","abstract_excerpt":"In this work we first show that the classical Thompson sampling algorithm for multi-arm bandits is differentially private as-is, without any modification. We provide per-round privacy guarantees as a function of problem parameters and show composition over $T$ rounds; since the algorithm is unchanged, existing $O(\\sqrt{NT\\log N})$ regret bounds still hold and there is no loss in performance due to privacy. We then show that simple modifications -- such as pre-pulling all arms a fixed number of times, increasing the sampling variance -- can provide tighter privacy guarantees. We again provide p"},"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":"2407.14879","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-20T14:01:03Z","cross_cats_sorted":["cs.DS","stat.ML"],"title_canon_sha256":"6c9e5198720313cfe970f65da385891439e5cd92c5532966929e061835d9b0ee","abstract_canon_sha256":"6fe84adca47f568399dab69ee6b1533a6d691ef997e9b46d1d12737522784fc0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:46:35.429192Z","signature_b64":"DpDQS1oIEKKh84b8Y7UrvbXkVtKCdGGoN/4U0NiM8OoD10NyYK9Rn4h9U1nvgIgNC2dhjutuaQ9+0F8cFKqaDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f9398ef726ac89212fd5b7c7065f7c2ab7ed59581c4cf7cdd418978014043b92","last_reissued_at":"2026-07-05T08:46:35.428743Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:46:35.428743Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Thompson Sampling Itself is Differentially Private","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DS","stat.ML"],"primary_cat":"cs.LG","authors_text":"Marco Avella Medina, Rachel Cummings, Tingting Ou","submitted_at":"2024-07-20T14:01:03Z","abstract_excerpt":"In this work we first show that the classical Thompson sampling algorithm for multi-arm bandits is differentially private as-is, without any modification. We provide per-round privacy guarantees as a function of problem parameters and show composition over $T$ rounds; since the algorithm is unchanged, existing $O(\\sqrt{NT\\log N})$ regret bounds still hold and there is no loss in performance due to privacy. We then show that simple modifications -- such as pre-pulling all arms a fixed number of times, increasing the sampling variance -- can provide tighter privacy guarantees. We again provide p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.14879","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/2407.14879/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":"2407.14879","created_at":"2026-07-05T08:46:35.428806+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.14879v1","created_at":"2026-07-05T08:46:35.428806+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.14879","created_at":"2026-07-05T08:46:35.428806+00:00"},{"alias_kind":"pith_short_12","alias_value":"7E4Y55ZGVSES","created_at":"2026-07-05T08:46:35.428806+00:00"},{"alias_kind":"pith_short_16","alias_value":"7E4Y55ZGVSESCL6V","created_at":"2026-07-05T08:46:35.428806+00:00"},{"alias_kind":"pith_short_8","alias_value":"7E4Y55ZG","created_at":"2026-07-05T08:46:35.428806+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.02383","citing_title":"Connecting Thompson Sampling and UCB: Towards More Efficient Trade-offs Between Privacy and Regret","ref_index":22,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7E4Y55ZGVSESCL6VW7DQMX34FK","json":"https://pith.science/pith/7E4Y55ZGVSESCL6VW7DQMX34FK.json","graph_json":"https://pith.science/api/pith-number/7E4Y55ZGVSESCL6VW7DQMX34FK/graph.json","events_json":"https://pith.science/api/pith-number/7E4Y55ZGVSESCL6VW7DQMX34FK/events.json","paper":"https://pith.science/paper/7E4Y55ZG"},"agent_actions":{"view_html":"https://pith.science/pith/7E4Y55ZGVSESCL6VW7DQMX34FK","download_json":"https://pith.science/pith/7E4Y55ZGVSESCL6VW7DQMX34FK.json","view_paper":"https://pith.science/paper/7E4Y55ZG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.14879&json=true","fetch_graph":"https://pith.science/api/pith-number/7E4Y55ZGVSESCL6VW7DQMX34FK/graph.json","fetch_events":"https://pith.science/api/pith-number/7E4Y55ZGVSESCL6VW7DQMX34FK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7E4Y55ZGVSESCL6VW7DQMX34FK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7E4Y55ZGVSESCL6VW7DQMX34FK/action/storage_attestation","attest_author":"https://pith.science/pith/7E4Y55ZGVSESCL6VW7DQMX34FK/action/author_attestation","sign_citation":"https://pith.science/pith/7E4Y55ZGVSESCL6VW7DQMX34FK/action/citation_signature","submit_replication":"https://pith.science/pith/7E4Y55ZGVSESCL6VW7DQMX34FK/action/replication_record"}},"created_at":"2026-07-05T08:46:35.428806+00:00","updated_at":"2026-07-05T08:46:35.428806+00:00"}