{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DFAT3OH26N7LOJDOLD62I6SFPW","short_pith_number":"pith:DFAT3OH2","schema_version":"1.0","canonical_sha256":"19413db8faf37eb7246e58fda47a457db9ea63f1b0d25584758e925c2bd341e2","source":{"kind":"arxiv","id":"2605.13426","version":2},"attestation_state":"computed","paper":{"title":"Strategic PAC Learnability via Geometric Definability","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A geometric definability assumption ensures strategic classification remains PAC learnable with sample complexity governed by formula complexity.","cross_cats":["math.AG"],"primary_cat":"cs.LG","authors_text":"Alexander Shlimovich, Elizaveta Nesterova, Nir Rosenfeld, Shay Moran, Yuval Filmus","submitted_at":"2026-05-13T12:21:56Z","abstract_excerpt":"Strategic classification studies learning settings in which individuals can modify their features, at a cost, in order to influence the classifier's decision. A central question is how the sample complexity of the induced (strategic) hypothesis class depends on the complexities of the underlying hypothesis class and the cost structure governing feasible manipulations. Prior work has shown that in several natural settings, such as linear classifiers with norm costs, the induced complexity can be controlled. We begin by showing that such guarantees fail in general - even in simple cases: there e"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2605.13426","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T12:21:56Z","cross_cats_sorted":["math.AG"],"title_canon_sha256":"8177de83887bb4d23e913e3b034f5a820e082352e521b45718ce1c5995d603db","abstract_canon_sha256":"1dff7a321cffcf6cfa146d4ac8313ad1b62769401446e2a186df49863503cd74"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:44:47.262667Z","signature_b64":"Cex9VCE9keg4uG3fYXkFI6S3R78xL9BF3PJljTy4JlfW2brss43tWwLE9lzixdMQVKVNG/8AoPv5sC+/GpdNCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"19413db8faf37eb7246e58fda47a457db9ea63f1b0d25584758e925c2bd341e2","last_reissued_at":"2026-05-18T02:44:47.262227Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:44:47.262227Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Strategic PAC Learnability via Geometric Definability","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A geometric definability assumption ensures strategic classification remains PAC learnable with sample complexity governed by formula complexity.","cross_cats":["math.AG"],"primary_cat":"cs.LG","authors_text":"Alexander Shlimovich, Elizaveta Nesterova, Nir Rosenfeld, Shay Moran, Yuval Filmus","submitted_at":"2026-05-13T12:21:56Z","abstract_excerpt":"Strategic classification studies learning settings in which individuals can modify their features, at a cost, in order to influence the classifier's decision. A central question is how the sample complexity of the induced (strategic) hypothesis class depends on the complexities of the underlying hypothesis class and the cost structure governing feasible manipulations. Prior work has shown that in several natural settings, such as linear classifiers with norm costs, the induced complexity can be controlled. We begin by showing that such guarantees fail in general - even in simple cases: there e"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Under the geometric definability assumption, learnability is preserved, with sample complexity controlled by the complexity of the defining formulas.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Both the hypothesis class and the cost-induced neighborhood relation can be defined by first-order formulas over R_exp.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Strategic PAC learnability is preserved when hypothesis classes and cost-induced neighborhoods are first-order definable over the reals with exponentiation, controlling sample complexity by formula complexity.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A geometric definability assumption ensures strategic classification remains PAC learnable with sample complexity governed by formula complexity.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"3ca2b499ca9843aca07eb51dd9bbb13a222ca4b39c97bc4e878c14a68b45c320"},"source":{"id":"2605.13426","kind":"arxiv","version":2},"verdict":{"id":"60cd19a9-7a9e-42b4-9244-cec2ef6ce5b2","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T05:18:21.035347Z","strongest_claim":"Under the geometric definability assumption, learnability is preserved, with sample complexity controlled by the complexity of the defining formulas.","one_line_summary":"Strategic PAC learnability is preserved when hypothesis classes and cost-induced neighborhoods are first-order definable over the reals with exponentiation, controlling sample complexity by formula complexity.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Both the hypothesis class and the cost-induced neighborhood relation can be defined by first-order formulas over R_exp.","pith_extraction_headline":"A geometric definability assumption ensures strategic classification remains PAC learnable with sample complexity governed by formula complexity."},"references":{"count":58,"sample":[{"doi":"","year":2013,"title":"Honest Compressions and Their Application to Compression Schemes , booktitle =","work_id":"911ed66c-fd71-423b-9eff-a83713231796","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.48550/arxiv.2312.00379","year":2023,"title":"CoRR , volume =","work_id":"1aeb54d6-cfa6-4cc9-9e6c-682d46db347c","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"Sign rank versus","work_id":"b4facebe-8088-4b71-99de-5971ceaf4611","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2002,"title":"Limitations of Learning Via Embeddings in Euclidean Half Spaces , journal =","work_id":"bc5dc28b-d0b1-430f-8b7b-1b76ebc2859f","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Talagrand , title =","work_id":"3669f16f-47fe-453c-9d05-526e3f2ffa2e","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":58,"snapshot_sha256":"c4f82e498924bd69fed7490ef2f85ac263b32209642b6952ed0bdad9356d0d69","internal_anchors":0},"formal_canon":{"evidence_count":3,"snapshot_sha256":"421392246b55b7d7ef5cc1993eb6ec3884aeb88f13b7989f64835767237f962c"},"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":"2605.13426","created_at":"2026-05-18T02:44:47.262294+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.13426v2","created_at":"2026-05-18T02:44:47.262294+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13426","created_at":"2026-05-18T02:44:47.262294+00:00"},{"alias_kind":"pith_short_12","alias_value":"DFAT3OH26N7L","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"DFAT3OH26N7LOJDO","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"DFAT3OH2","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":3,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DFAT3OH26N7LOJDOLD62I6SFPW","json":"https://pith.science/pith/DFAT3OH26N7LOJDOLD62I6SFPW.json","graph_json":"https://pith.science/api/pith-number/DFAT3OH26N7LOJDOLD62I6SFPW/graph.json","events_json":"https://pith.science/api/pith-number/DFAT3OH26N7LOJDOLD62I6SFPW/events.json","paper":"https://pith.science/paper/DFAT3OH2"},"agent_actions":{"view_html":"https://pith.science/pith/DFAT3OH26N7LOJDOLD62I6SFPW","download_json":"https://pith.science/pith/DFAT3OH26N7LOJDOLD62I6SFPW.json","view_paper":"https://pith.science/paper/DFAT3OH2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.13426&json=true","fetch_graph":"https://pith.science/api/pith-number/DFAT3OH26N7LOJDOLD62I6SFPW/graph.json","fetch_events":"https://pith.science/api/pith-number/DFAT3OH26N7LOJDOLD62I6SFPW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DFAT3OH26N7LOJDOLD62I6SFPW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DFAT3OH26N7LOJDOLD62I6SFPW/action/storage_attestation","attest_author":"https://pith.science/pith/DFAT3OH26N7LOJDOLD62I6SFPW/action/author_attestation","sign_citation":"https://pith.science/pith/DFAT3OH26N7LOJDOLD62I6SFPW/action/citation_signature","submit_replication":"https://pith.science/pith/DFAT3OH26N7LOJDOLD62I6SFPW/action/replication_record"}},"created_at":"2026-05-18T02:44:47.262294+00:00","updated_at":"2026-05-18T02:44:47.262294+00:00"}