{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:223YGI7WNLL2B2IMK6FYEQR36T","short_pith_number":"pith:223YGI7W","schema_version":"1.0","canonical_sha256":"d6b78323f66ad7a0e90c578b82423bf4e35a0d852828de31138925d934e3dcf2","source":{"kind":"arxiv","id":"2409.13953","version":1},"attestation_state":"computed","paper":{"title":"Training Large ASR Encoders with Differential Privacy","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.CR","cs.LG","eess.AS"],"primary_cat":"cs.SD","authors_text":"Abhradeep Thakurta, Arun Narayanan, Geeticka Chauhan, Om Thakkar, Steve Chien","submitted_at":"2024-09-21T00:01:49Z","abstract_excerpt":"Self-supervised learning (SSL) methods for large speech models have proven to be highly effective at ASR. With the interest in public deployment of large pre-trained models, there is a rising concern for unintended memorization and leakage of sensitive data points from the training data. In this paper, we apply differentially private (DP) pre-training to a SOTA Conformer-based encoder, and study its performance on a downstream ASR task assuming the fine-tuning data is public. This paper is the first to apply DP to SSL for ASR, investigating the DP noise tolerance of the BEST-RQ pre-training me"},"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":"2409.13953","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.SD","submitted_at":"2024-09-21T00:01:49Z","cross_cats_sorted":["cs.CR","cs.LG","eess.AS"],"title_canon_sha256":"32b18ac995597d01aaaef21966165ba381ad91e838c0e9326491d6ffdb092243","abstract_canon_sha256":"3cf9e4bbe906571f0206721668c8d2d34d6e0c51cf00fa35ac8f5da5a1d6f150"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:10:09.275874Z","signature_b64":"shcWNQtdWC118KN6qRyf4LArtmaqrA/2VT5ZFl6tPfHZ+MazXdxTZbD5cOmb/P+DechxyQsN7paByAu3RGWTDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d6b78323f66ad7a0e90c578b82423bf4e35a0d852828de31138925d934e3dcf2","last_reissued_at":"2026-07-05T09:10:09.275393Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:10:09.275393Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Training Large ASR Encoders with Differential Privacy","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.CR","cs.LG","eess.AS"],"primary_cat":"cs.SD","authors_text":"Abhradeep Thakurta, Arun Narayanan, Geeticka Chauhan, Om Thakkar, Steve Chien","submitted_at":"2024-09-21T00:01:49Z","abstract_excerpt":"Self-supervised learning (SSL) methods for large speech models have proven to be highly effective at ASR. With the interest in public deployment of large pre-trained models, there is a rising concern for unintended memorization and leakage of sensitive data points from the training data. In this paper, we apply differentially private (DP) pre-training to a SOTA Conformer-based encoder, and study its performance on a downstream ASR task assuming the fine-tuning data is public. This paper is the first to apply DP to SSL for ASR, investigating the DP noise tolerance of the BEST-RQ pre-training me"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.13953","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/2409.13953/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":"2409.13953","created_at":"2026-07-05T09:10:09.275452+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.13953v1","created_at":"2026-07-05T09:10:09.275452+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.13953","created_at":"2026-07-05T09:10:09.275452+00:00"},{"alias_kind":"pith_short_12","alias_value":"223YGI7WNLL2","created_at":"2026-07-05T09:10:09.275452+00:00"},{"alias_kind":"pith_short_16","alias_value":"223YGI7WNLL2B2IM","created_at":"2026-07-05T09:10:09.275452+00:00"},{"alias_kind":"pith_short_8","alias_value":"223YGI7W","created_at":"2026-07-05T09:10:09.275452+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.00451","citing_title":"Towards Privacy-aware Mental Health AI Models: Advances, Challenges, and Opportunities","ref_index":43,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/223YGI7WNLL2B2IMK6FYEQR36T","json":"https://pith.science/pith/223YGI7WNLL2B2IMK6FYEQR36T.json","graph_json":"https://pith.science/api/pith-number/223YGI7WNLL2B2IMK6FYEQR36T/graph.json","events_json":"https://pith.science/api/pith-number/223YGI7WNLL2B2IMK6FYEQR36T/events.json","paper":"https://pith.science/paper/223YGI7W"},"agent_actions":{"view_html":"https://pith.science/pith/223YGI7WNLL2B2IMK6FYEQR36T","download_json":"https://pith.science/pith/223YGI7WNLL2B2IMK6FYEQR36T.json","view_paper":"https://pith.science/paper/223YGI7W","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.13953&json=true","fetch_graph":"https://pith.science/api/pith-number/223YGI7WNLL2B2IMK6FYEQR36T/graph.json","fetch_events":"https://pith.science/api/pith-number/223YGI7WNLL2B2IMK6FYEQR36T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/223YGI7WNLL2B2IMK6FYEQR36T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/223YGI7WNLL2B2IMK6FYEQR36T/action/storage_attestation","attest_author":"https://pith.science/pith/223YGI7WNLL2B2IMK6FYEQR36T/action/author_attestation","sign_citation":"https://pith.science/pith/223YGI7WNLL2B2IMK6FYEQR36T/action/citation_signature","submit_replication":"https://pith.science/pith/223YGI7WNLL2B2IMK6FYEQR36T/action/replication_record"}},"created_at":"2026-07-05T09:10:09.275452+00:00","updated_at":"2026-07-05T09:10:09.275452+00:00"}