{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:FYH2GCCUGLH3UOFJMFQT4BPKY2","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"929a29e16bb21d11328f993c7140eaf10cd3029d6de0381f2aa5b22165d11321","cross_cats_sorted":["cs.IT","math.IT"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-28T10:46:44Z","title_canon_sha256":"9fe0143d5a93dd39115324fefc5a76f4a694b79322ec50248e7309f48ea00025"},"schema_version":"1.0","source":{"id":"2601.20477","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2601.20477","created_at":"2026-05-20T00:01:39Z"},{"alias_kind":"arxiv_version","alias_value":"2601.20477v4","created_at":"2026-05-20T00:01:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.20477","created_at":"2026-05-20T00:01:39Z"},{"alias_kind":"pith_short_12","alias_value":"FYH2GCCUGLH3","created_at":"2026-05-20T00:01:39Z"},{"alias_kind":"pith_short_16","alias_value":"FYH2GCCUGLH3UOFJ","created_at":"2026-05-20T00:01:39Z"},{"alias_kind":"pith_short_8","alias_value":"FYH2GCCU","created_at":"2026-05-20T00:01:39Z"}],"graph_snapshots":[{"event_id":"sha256:13304e94f96cd44e5eaf277874797bc1b6dbbdb33b31ce40ff157b290cfea736","target":"graph","created_at":"2026-05-20T00:01:39Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"along training trajectories, well-generalizing networks progressively approach Neyman-Pearson optimal decision rules, as measured by monotonic growth in the KL divergence retained by learned representations"},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the class-conditional distributions induced by the learned representations are sufficiently well-behaved for the binary hypothesis tests to be meaningfully defined and that monotonic KL growth directly indicates approach to Neyman-Pearson optimality without additional assumptions on the data distribution or network architecture."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Well-generalizing neural classifiers approach Neyman-Pearson optimal rules as training proceeds, shown by monotonic increase in KL divergence preserved in their representations."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Neural networks approach Neyman-Pearson optimal rules through monotonic growth in retained KL divergence."}],"snapshot_sha256":"d356a1a645c30626b98129927f9a328f3109cfd56c27a327e55e392778d910a0"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"97b32db9989626cd8d4700ad5d056f5ce8880c5ab8bd603be45bb6108e713979"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2601.20477/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We study the training dynamics of neural classifiers through the lens of binary hypothesis testing. We re-formalize classification as a collection of binary tests between class-conditional distributions induced by learned representations and show empirically that, along training trajectories, well-generalizing networks progressively approach Neyman-Pearson optimal decision rules, as measured by monotonic growth in the KL divergence retained by learned representations. We provide sufficient conditions for exact optimality, discuss its implications for training regularization, and define an info","authors_text":"Kadircan Aksoy, Peter Jung, Protim Bhattacharjee","cross_cats":["cs.IT","math.IT"],"headline":"Neural networks approach Neyman-Pearson optimal rules through monotonic growth in retained KL divergence.","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-28T10:46:44Z","title":"Implicit Hypothesis Testing and Divergence Preservation in Neural Network Representations"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.20477","kind":"arxiv","version":4},"verdict":{"created_at":"2026-05-16T10:48:01.305553Z","id":"a56fb9b5-e34d-4421-8e68-a09af1b8defd","model_set":{"reader":"grok-4.3"},"one_line_summary":"Well-generalizing neural classifiers approach Neyman-Pearson optimal rules as training proceeds, shown by monotonic increase in KL divergence preserved in their representations.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Neural networks approach Neyman-Pearson optimal rules through monotonic growth in retained KL divergence.","strongest_claim":"along training trajectories, well-generalizing networks progressively approach Neyman-Pearson optimal decision rules, as measured by monotonic growth in the KL divergence retained by learned representations","weakest_assumption":"That the class-conditional distributions induced by the learned representations are sufficiently well-behaved for the binary hypothesis tests to be meaningfully defined and that monotonic KL growth directly indicates approach to Neyman-Pearson optimality without additional assumptions on the data distribution or network architecture."}},"verdict_id":"a56fb9b5-e34d-4421-8e68-a09af1b8defd"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:af25485a405af23f6c5c98280676d072c97ffb44911bb6661b4c9feeabcd8d6d","target":"record","created_at":"2026-05-20T00:01:39Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"929a29e16bb21d11328f993c7140eaf10cd3029d6de0381f2aa5b22165d11321","cross_cats_sorted":["cs.IT","math.IT"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-28T10:46:44Z","title_canon_sha256":"9fe0143d5a93dd39115324fefc5a76f4a694b79322ec50248e7309f48ea00025"},"schema_version":"1.0","source":{"id":"2601.20477","kind":"arxiv","version":4}},"canonical_sha256":"2e0fa3085432cfba38a961613e05eac68fd38dea5a6a32e78cba60e1c1e55aef","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2e0fa3085432cfba38a961613e05eac68fd38dea5a6a32e78cba60e1c1e55aef","first_computed_at":"2026-05-20T00:01:39.323841Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:01:39.323841Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ORWmwsO43CM6VJ4xJkXxvUQhY264dNq3fPx8gPHsN9i1+rE1U9wSnEYCRYr0i0ZaFVORTOvW/onES4bBEqFODA==","signature_status":"signed_v1","signed_at":"2026-05-20T00:01:39.324586Z","signed_message":"canonical_sha256_bytes"},"source_id":"2601.20477","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:af25485a405af23f6c5c98280676d072c97ffb44911bb6661b4c9feeabcd8d6d","sha256:13304e94f96cd44e5eaf277874797bc1b6dbbdb33b31ce40ff157b290cfea736"],"state_sha256":"01dea63871e3ce7fdbea6a32d80c632e24e430be2cfe849e5e442956e9457b7a"}