{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:BFKO6DMD5XJFQUPD4DTAYAJKNK","short_pith_number":"pith:BFKO6DMD","schema_version":"1.0","canonical_sha256":"0954ef0d83edd25851e3e0e60c012a6a906c9e01b91dc2e0f32aeaf35ea0a107","source":{"kind":"arxiv","id":"2605.26222","version":1},"attestation_state":"computed","paper":{"title":"From Privacy to Generalization: Linear Max-Information Bounds for DP-SGD","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Christoph H. Lampert, Hossein Zakerinia","submitted_at":"2026-05-25T18:00:05Z","abstract_excerpt":"Understanding the relationship between generalization and privacy remains a central challenge in modern machine learning theory, particularly for deep networks trained by variants of differentially private stochastic gradient descent (DP-SGD). In this work we make progress on this persistent open problem by proving a finite-sample bound on the approximate max-information of DP-SGD that exhibits scaling properties comparable with (Dwork et al, 2015)'s classic result for $\\epsilon$-differentially private algorithms, namely at most linear in the dataset size. From our result we obtain a general-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":"2605.26222","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-25T18:00:05Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"3bbfc036715660275eb1e6596a7795f5b86839daa6509385a18e9eaabb0c5ea9","abstract_canon_sha256":"46f5481476a6e3b9015a65bc0f798f35565a72f6fb34ac95cc7532d5143beb1f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T00:04:29.488541Z","signature_b64":"m21Ay+yLtqcknquOTAZBPUQmWpRnZ0NqBNPkb3N5kUoRedwWswdCD1fHFDUJmCCBds+bKRwgJcCEtV53n8WfBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0954ef0d83edd25851e3e0e60c012a6a906c9e01b91dc2e0f32aeaf35ea0a107","last_reissued_at":"2026-05-27T00:04:29.487845Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T00:04:29.487845Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"From Privacy to Generalization: Linear Max-Information Bounds for DP-SGD","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Christoph H. Lampert, Hossein Zakerinia","submitted_at":"2026-05-25T18:00:05Z","abstract_excerpt":"Understanding the relationship between generalization and privacy remains a central challenge in modern machine learning theory, particularly for deep networks trained by variants of differentially private stochastic gradient descent (DP-SGD). In this work we make progress on this persistent open problem by proving a finite-sample bound on the approximate max-information of DP-SGD that exhibits scaling properties comparable with (Dwork et al, 2015)'s classic result for $\\epsilon$-differentially private algorithms, namely at most linear in the dataset size. From our result we obtain a general-p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26222","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/2605.26222/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":"2605.26222","created_at":"2026-05-27T00:04:29.487948+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.26222v1","created_at":"2026-05-27T00:04:29.487948+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26222","created_at":"2026-05-27T00:04:29.487948+00:00"},{"alias_kind":"pith_short_12","alias_value":"BFKO6DMD5XJF","created_at":"2026-05-27T00:04:29.487948+00:00"},{"alias_kind":"pith_short_16","alias_value":"BFKO6DMD5XJFQUPD","created_at":"2026-05-27T00:04:29.487948+00:00"},{"alias_kind":"pith_short_8","alias_value":"BFKO6DMD","created_at":"2026-05-27T00:04:29.487948+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/BFKO6DMD5XJFQUPD4DTAYAJKNK","json":"https://pith.science/pith/BFKO6DMD5XJFQUPD4DTAYAJKNK.json","graph_json":"https://pith.science/api/pith-number/BFKO6DMD5XJFQUPD4DTAYAJKNK/graph.json","events_json":"https://pith.science/api/pith-number/BFKO6DMD5XJFQUPD4DTAYAJKNK/events.json","paper":"https://pith.science/paper/BFKO6DMD"},"agent_actions":{"view_html":"https://pith.science/pith/BFKO6DMD5XJFQUPD4DTAYAJKNK","download_json":"https://pith.science/pith/BFKO6DMD5XJFQUPD4DTAYAJKNK.json","view_paper":"https://pith.science/paper/BFKO6DMD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.26222&json=true","fetch_graph":"https://pith.science/api/pith-number/BFKO6DMD5XJFQUPD4DTAYAJKNK/graph.json","fetch_events":"https://pith.science/api/pith-number/BFKO6DMD5XJFQUPD4DTAYAJKNK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BFKO6DMD5XJFQUPD4DTAYAJKNK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BFKO6DMD5XJFQUPD4DTAYAJKNK/action/storage_attestation","attest_author":"https://pith.science/pith/BFKO6DMD5XJFQUPD4DTAYAJKNK/action/author_attestation","sign_citation":"https://pith.science/pith/BFKO6DMD5XJFQUPD4DTAYAJKNK/action/citation_signature","submit_replication":"https://pith.science/pith/BFKO6DMD5XJFQUPD4DTAYAJKNK/action/replication_record"}},"created_at":"2026-05-27T00:04:29.487948+00:00","updated_at":"2026-05-27T00:04:29.487948+00:00"}