{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:3RAFHSEXNBVBMAWDQL6LVA6E5C","short_pith_number":"pith:3RAFHSEX","schema_version":"1.0","canonical_sha256":"dc4053c897686a1602c382fcba83c4e894df55a33e4c6375ae7ae059eda45394","source":{"kind":"arxiv","id":"2505.06169","version":1},"attestation_state":"computed","paper":{"title":"On the Depth of Monotone ReLU Neural Networks and ICNNs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.DM","cs.NE","math.CO"],"primary_cat":"cs.LG","authors_text":"Amir Yehudayoff, Christoph Hertrich, Daniel Reichman, Egor Bakaev, Florestan Brunck","submitted_at":"2025-05-09T16:19:34Z","abstract_excerpt":"We study two models of ReLU neural networks: monotone networks (ReLU$^+$) and input convex neural networks (ICNN). Our focus is on expressivity, mostly in terms of depth, and we prove the following lower bounds. For the maximum function MAX$_n$ computing the maximum of $n$ real numbers, we show that ReLU$^+$ networks cannot compute MAX$_n$, or even approximate it. We prove a sharp $n$ lower bound on the ICNN depth complexity of MAX$_n$. We also prove depth separations between ReLU networks and ICNNs; for every $k$, there is a depth-2 ReLU network of size $O(k^2)$ that cannot be simulated by a "},"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":"2505.06169","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-09T16:19:34Z","cross_cats_sorted":["cs.DM","cs.NE","math.CO"],"title_canon_sha256":"c81e6b1f4ec8f3122ca699b54c2288801b800ceec57a41ffa39a9174f96f10a1","abstract_canon_sha256":"ab78d4d11fe6e096e949cff0798ee6e85a5847351626e1a997b2e8bfb00c1a7e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:00:55.279296Z","signature_b64":"j0eoaNhI3/84yy92wahOaodETj9zpsPNdKmCq/1m41puzaUDU6zoFQN2F58FxomIDm79PQaQxNdl1Wtrty05Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dc4053c897686a1602c382fcba83c4e894df55a33e4c6375ae7ae059eda45394","last_reissued_at":"2026-07-05T11:00:55.278797Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:00:55.278797Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Depth of Monotone ReLU Neural Networks and ICNNs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.DM","cs.NE","math.CO"],"primary_cat":"cs.LG","authors_text":"Amir Yehudayoff, Christoph Hertrich, Daniel Reichman, Egor Bakaev, Florestan Brunck","submitted_at":"2025-05-09T16:19:34Z","abstract_excerpt":"We study two models of ReLU neural networks: monotone networks (ReLU$^+$) and input convex neural networks (ICNN). Our focus is on expressivity, mostly in terms of depth, and we prove the following lower bounds. For the maximum function MAX$_n$ computing the maximum of $n$ real numbers, we show that ReLU$^+$ networks cannot compute MAX$_n$, or even approximate it. We prove a sharp $n$ lower bound on the ICNN depth complexity of MAX$_n$. We also prove depth separations between ReLU networks and ICNNs; for every $k$, there is a depth-2 ReLU network of size $O(k^2)$ that cannot be simulated by a "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.06169","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/2505.06169/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":"2505.06169","created_at":"2026-07-05T11:00:55.278856+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.06169v1","created_at":"2026-07-05T11:00:55.278856+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.06169","created_at":"2026-07-05T11:00:55.278856+00:00"},{"alias_kind":"pith_short_12","alias_value":"3RAFHSEXNBVB","created_at":"2026-07-05T11:00:55.278856+00:00"},{"alias_kind":"pith_short_16","alias_value":"3RAFHSEXNBVBMAWD","created_at":"2026-07-05T11:00:55.278856+00:00"},{"alias_kind":"pith_short_8","alias_value":"3RAFHSEX","created_at":"2026-07-05T11:00:55.278856+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.22355","citing_title":"SOC-ICNN: From Polyhedral to Conic Geometry for Learning Convex Surrogate Functions","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3RAFHSEXNBVBMAWDQL6LVA6E5C","json":"https://pith.science/pith/3RAFHSEXNBVBMAWDQL6LVA6E5C.json","graph_json":"https://pith.science/api/pith-number/3RAFHSEXNBVBMAWDQL6LVA6E5C/graph.json","events_json":"https://pith.science/api/pith-number/3RAFHSEXNBVBMAWDQL6LVA6E5C/events.json","paper":"https://pith.science/paper/3RAFHSEX"},"agent_actions":{"view_html":"https://pith.science/pith/3RAFHSEXNBVBMAWDQL6LVA6E5C","download_json":"https://pith.science/pith/3RAFHSEXNBVBMAWDQL6LVA6E5C.json","view_paper":"https://pith.science/paper/3RAFHSEX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.06169&json=true","fetch_graph":"https://pith.science/api/pith-number/3RAFHSEXNBVBMAWDQL6LVA6E5C/graph.json","fetch_events":"https://pith.science/api/pith-number/3RAFHSEXNBVBMAWDQL6LVA6E5C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3RAFHSEXNBVBMAWDQL6LVA6E5C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3RAFHSEXNBVBMAWDQL6LVA6E5C/action/storage_attestation","attest_author":"https://pith.science/pith/3RAFHSEXNBVBMAWDQL6LVA6E5C/action/author_attestation","sign_citation":"https://pith.science/pith/3RAFHSEXNBVBMAWDQL6LVA6E5C/action/citation_signature","submit_replication":"https://pith.science/pith/3RAFHSEXNBVBMAWDQL6LVA6E5C/action/replication_record"}},"created_at":"2026-07-05T11:00:55.278856+00:00","updated_at":"2026-07-05T11:00:55.278856+00:00"}