{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:GBTDWRSZ3GIR2TLJRF2ZDMXRUW","short_pith_number":"pith:GBTDWRSZ","schema_version":"1.0","canonical_sha256":"30663b4659d9911d4d69897591b2f1a59c85309f32d7e1a97d003c57cd3e5c65","source":{"kind":"arxiv","id":"2404.11262","version":1},"attestation_state":"computed","paper":{"title":"Sampling-based Pseudo-Likelihood for Membership Inference Attacks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Masahiro Kaneko, Naoaki Okazaki, Youmi Ma, Yuki Wata","submitted_at":"2024-04-17T11:12:59Z","abstract_excerpt":"Large Language Models (LLMs) are trained on large-scale web data, which makes it difficult to grasp the contribution of each text. This poses the risk of leaking inappropriate data such as benchmarks, personal information, and copyrighted texts in the training data. Membership Inference Attacks (MIA), which determine whether a given text is included in the model's training data, have been attracting attention. Previous studies of MIAs revealed that likelihood-based classification is effective for detecting leaks in LLMs. However, the existing methods cannot be applied to some proprietary model"},"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":"2404.11262","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-04-17T11:12:59Z","cross_cats_sorted":[],"title_canon_sha256":"e6f39a46677d7c5db5bb0be61d6dfabb7f87698130825080a413d188935c2339","abstract_canon_sha256":"4441ce77cec7934be4d74b5939428a0f894f88c8565b580c394bea56eae64d26"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:09:06.616375Z","signature_b64":"8m+W2DecE4k8u3vGL6KcY5AHZqBxSDnX5uJ7k90M3aC1Kc4vnH+WdiTdRb4IBhjeE8ZoKNE7/LEHbM35OQb7Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"30663b4659d9911d4d69897591b2f1a59c85309f32d7e1a97d003c57cd3e5c65","last_reissued_at":"2026-07-05T08:09:06.615860Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:09:06.615860Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sampling-based Pseudo-Likelihood for Membership Inference Attacks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Masahiro Kaneko, Naoaki Okazaki, Youmi Ma, Yuki Wata","submitted_at":"2024-04-17T11:12:59Z","abstract_excerpt":"Large Language Models (LLMs) are trained on large-scale web data, which makes it difficult to grasp the contribution of each text. This poses the risk of leaking inappropriate data such as benchmarks, personal information, and copyrighted texts in the training data. Membership Inference Attacks (MIA), which determine whether a given text is included in the model's training data, have been attracting attention. Previous studies of MIAs revealed that likelihood-based classification is effective for detecting leaks in LLMs. However, the existing methods cannot be applied to some proprietary model"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.11262","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/2404.11262/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":"2404.11262","created_at":"2026-07-05T08:09:06.615923+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.11262v1","created_at":"2026-07-05T08:09:06.615923+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.11262","created_at":"2026-07-05T08:09:06.615923+00:00"},{"alias_kind":"pith_short_12","alias_value":"GBTDWRSZ3GIR","created_at":"2026-07-05T08:09:06.615923+00:00"},{"alias_kind":"pith_short_16","alias_value":"GBTDWRSZ3GIR2TLJ","created_at":"2026-07-05T08:09:06.615923+00:00"},{"alias_kind":"pith_short_8","alias_value":"GBTDWRSZ","created_at":"2026-07-05T08:09:06.615923+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.26133","citing_title":"Pretraining Data Exposure in Large Language Models: A Survey of Membership Inference, Data Contamination, and Security Implications","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GBTDWRSZ3GIR2TLJRF2ZDMXRUW","json":"https://pith.science/pith/GBTDWRSZ3GIR2TLJRF2ZDMXRUW.json","graph_json":"https://pith.science/api/pith-number/GBTDWRSZ3GIR2TLJRF2ZDMXRUW/graph.json","events_json":"https://pith.science/api/pith-number/GBTDWRSZ3GIR2TLJRF2ZDMXRUW/events.json","paper":"https://pith.science/paper/GBTDWRSZ"},"agent_actions":{"view_html":"https://pith.science/pith/GBTDWRSZ3GIR2TLJRF2ZDMXRUW","download_json":"https://pith.science/pith/GBTDWRSZ3GIR2TLJRF2ZDMXRUW.json","view_paper":"https://pith.science/paper/GBTDWRSZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.11262&json=true","fetch_graph":"https://pith.science/api/pith-number/GBTDWRSZ3GIR2TLJRF2ZDMXRUW/graph.json","fetch_events":"https://pith.science/api/pith-number/GBTDWRSZ3GIR2TLJRF2ZDMXRUW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GBTDWRSZ3GIR2TLJRF2ZDMXRUW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GBTDWRSZ3GIR2TLJRF2ZDMXRUW/action/storage_attestation","attest_author":"https://pith.science/pith/GBTDWRSZ3GIR2TLJRF2ZDMXRUW/action/author_attestation","sign_citation":"https://pith.science/pith/GBTDWRSZ3GIR2TLJRF2ZDMXRUW/action/citation_signature","submit_replication":"https://pith.science/pith/GBTDWRSZ3GIR2TLJRF2ZDMXRUW/action/replication_record"}},"created_at":"2026-07-05T08:09:06.615923+00:00","updated_at":"2026-07-05T08:09:06.615923+00:00"}