{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:OWPK4BFY47CJHQVEV4BYDH6Z7D","short_pith_number":"pith:OWPK4BFY","canonical_record":{"source":{"id":"2602.07618","kind":"arxiv","version":6},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-07T16:52:38Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"400bd5230f4c4ac176391aeb223bfe9a66d02d7e527c151b9cddd0ace5081e39","abstract_canon_sha256":"d401ffe1572aa2b0b91d614c54be15dab4df12a69967d262d644a3b69113f001"},"schema_version":"1.0"},"canonical_sha256":"759eae04b8e7c493c2a4af03819fd9f8e792acd6b7cdd40d4aea29745088d97d","source":{"kind":"arxiv","id":"2602.07618","version":6},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.07618","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"arxiv_version","alias_value":"2602.07618v6","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.07618","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"pith_short_12","alias_value":"OWPK4BFY47CJ","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"pith_short_16","alias_value":"OWPK4BFY47CJHQVE","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"pith_short_8","alias_value":"OWPK4BFY","created_at":"2026-05-20T00:05:41Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:OWPK4BFY47CJHQVEV4BYDH6Z7D","target":"record","payload":{"canonical_record":{"source":{"id":"2602.07618","kind":"arxiv","version":6},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-07T16:52:38Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"400bd5230f4c4ac176391aeb223bfe9a66d02d7e527c151b9cddd0ace5081e39","abstract_canon_sha256":"d401ffe1572aa2b0b91d614c54be15dab4df12a69967d262d644a3b69113f001"},"schema_version":"1.0"},"canonical_sha256":"759eae04b8e7c493c2a4af03819fd9f8e792acd6b7cdd40d4aea29745088d97d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:05:41.791117Z","signature_b64":"nBmBirZpD7oKtVA3+lJb3o5I1vIPpf0zcBPfp1j5zIGKf7+yFo3sMSJUxfAj/Jp41QPFaEsb3hLC+ZX/SsJ/CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"759eae04b8e7c493c2a4af03819fd9f8e792acd6b7cdd40d4aea29745088d97d","last_reissued_at":"2026-05-20T00:05:41.790668Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:05:41.790668Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2602.07618","source_version":6,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:05:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"sOkheTqmGdRdWa6mxTk/FNc0yYk2+S6/6MCfLHxcXlW1harw70DxNrziHnt5Y9A5+9AonX0vLqw+hE3qCBFKDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T18:42:42.145538Z"},"content_sha256":"a281a6323011335accb87f7241de26396bf4848cc181ee5fe6897affa76d9ad7","schema_version":"1.0","event_id":"sha256:a281a6323011335accb87f7241de26396bf4848cc181ee5fe6897affa76d9ad7"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:OWPK4BFY47CJHQVEV4BYDH6Z7D","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Neural Networks With Dense Weights Are Not Universal Approximators","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Dense neural networks with constraints on weights and dimensions cannot approximate all Lipschitz continuous functions.","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Levi Rauchwerger, Ron Levie, Stefanie Jegelka","submitted_at":"2026-02-07T16:52:38Z","abstract_excerpt":"We investigate the approximation capabilities of dense neural networks. While universal approximation theorems establish that sufficiently large architectures can approximate arbitrary continuous functions if there are no restrictions on the weight values, we show that dense neural networks do not possess this universality. Our argument is based on a model compression approach, combining the weak regularity lemma with an interpretation of feedforward networks as message passing graph neural networks. We consider ReLU neural networks subject to natural constraints on weights and input and outpu"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We demonstrate the existence of Lipschitz continuous functions that cannot be approximated by such networks.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The chosen constraints on weights and dimensions accurately model a notion of dense connectivity, and feedforward networks can be interpreted as message passing graph neural networks without loss of generality for the approximation question.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Dense neural networks subject to constraints on weights and dimensions cannot approximate arbitrary Lipschitz continuous functions.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Dense neural networks with constraints on weights and dimensions cannot approximate all Lipschitz continuous functions.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"bad81cd79800ef872b814e79da31ef80091724e017ac9e4dc313d691ac316e2d"},"source":{"id":"2602.07618","kind":"arxiv","version":6},"verdict":{"id":"b26cf5a3-c12a-4dd6-90a1-3f941da48302","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T06:01:51.254717Z","strongest_claim":"We demonstrate the existence of Lipschitz continuous functions that cannot be approximated by such networks.","one_line_summary":"Dense neural networks subject to constraints on weights and dimensions cannot approximate arbitrary Lipschitz continuous functions.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The chosen constraints on weights and dimensions accurately model a notion of dense connectivity, and feedforward networks can be interpreted as message passing graph neural networks without loss of generality for the approximation question.","pith_extraction_headline":"Dense neural networks with constraints on weights and dimensions cannot approximate all Lipschitz continuous functions."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2602.07618/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":2,"snapshot_sha256":"3e5b11c194875d39632d3341adc759d2e13b9399301a0abb89b365775751a9a4"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"b26cf5a3-c12a-4dd6-90a1-3f941da48302"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:05:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"+QCxpjbcsqfPWGFVN9UdYhTJR0OWnoXUcM4msUcv7OTaz4tVb3uEYdz3R3hR6fZRE+EFccJ8/V/BlfNXD1j2Bg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T18:42:42.146446Z"},"content_sha256":"d30b2e2f7e661c4b2164a8509e07b28ed21a653fc3c2ab5ace1fbc64e92bdd7d","schema_version":"1.0","event_id":"sha256:d30b2e2f7e661c4b2164a8509e07b28ed21a653fc3c2ab5ace1fbc64e92bdd7d"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/OWPK4BFY47CJHQVEV4BYDH6Z7D/bundle.json","state_url":"https://pith.science/pith/OWPK4BFY47CJHQVEV4BYDH6Z7D/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/OWPK4BFY47CJHQVEV4BYDH6Z7D/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-27T18:42:42Z","links":{"resolver":"https://pith.science/pith/OWPK4BFY47CJHQVEV4BYDH6Z7D","bundle":"https://pith.science/pith/OWPK4BFY47CJHQVEV4BYDH6Z7D/bundle.json","state":"https://pith.science/pith/OWPK4BFY47CJHQVEV4BYDH6Z7D/state.json","well_known_bundle":"https://pith.science/.well-known/pith/OWPK4BFY47CJHQVEV4BYDH6Z7D/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:OWPK4BFY47CJHQVEV4BYDH6Z7D","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":"d401ffe1572aa2b0b91d614c54be15dab4df12a69967d262d644a3b69113f001","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-07T16:52:38Z","title_canon_sha256":"400bd5230f4c4ac176391aeb223bfe9a66d02d7e527c151b9cddd0ace5081e39"},"schema_version":"1.0","source":{"id":"2602.07618","kind":"arxiv","version":6}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.07618","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"arxiv_version","alias_value":"2602.07618v6","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.07618","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"pith_short_12","alias_value":"OWPK4BFY47CJ","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"pith_short_16","alias_value":"OWPK4BFY47CJHQVE","created_at":"2026-05-20T00:05:41Z"},{"alias_kind":"pith_short_8","alias_value":"OWPK4BFY","created_at":"2026-05-20T00:05:41Z"}],"graph_snapshots":[{"event_id":"sha256:d30b2e2f7e661c4b2164a8509e07b28ed21a653fc3c2ab5ace1fbc64e92bdd7d","target":"graph","created_at":"2026-05-20T00:05:41Z","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":"We demonstrate the existence of Lipschitz continuous functions that cannot be approximated by such networks."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The chosen constraints on weights and dimensions accurately model a notion of dense connectivity, and feedforward networks can be interpreted as message passing graph neural networks without loss of generality for the approximation question."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Dense neural networks subject to constraints on weights and dimensions cannot approximate arbitrary Lipschitz continuous functions."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Dense neural networks with constraints on weights and dimensions cannot approximate all Lipschitz continuous functions."}],"snapshot_sha256":"bad81cd79800ef872b814e79da31ef80091724e017ac9e4dc313d691ac316e2d"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"3e5b11c194875d39632d3341adc759d2e13b9399301a0abb89b365775751a9a4"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2602.07618/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We investigate the approximation capabilities of dense neural networks. While universal approximation theorems establish that sufficiently large architectures can approximate arbitrary continuous functions if there are no restrictions on the weight values, we show that dense neural networks do not possess this universality. Our argument is based on a model compression approach, combining the weak regularity lemma with an interpretation of feedforward networks as message passing graph neural networks. We consider ReLU neural networks subject to natural constraints on weights and input and outpu","authors_text":"Levi Rauchwerger, Ron Levie, Stefanie Jegelka","cross_cats":["stat.ML"],"headline":"Dense neural networks with constraints on weights and dimensions cannot approximate all Lipschitz continuous functions.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-07T16:52:38Z","title":"Neural Networks With Dense Weights Are Not Universal Approximators"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.07618","kind":"arxiv","version":6},"verdict":{"created_at":"2026-05-16T06:01:51.254717Z","id":"b26cf5a3-c12a-4dd6-90a1-3f941da48302","model_set":{"reader":"grok-4.3"},"one_line_summary":"Dense neural networks subject to constraints on weights and dimensions cannot approximate arbitrary Lipschitz continuous functions.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Dense neural networks with constraints on weights and dimensions cannot approximate all Lipschitz continuous functions.","strongest_claim":"We demonstrate the existence of Lipschitz continuous functions that cannot be approximated by such networks.","weakest_assumption":"The chosen constraints on weights and dimensions accurately model a notion of dense connectivity, and feedforward networks can be interpreted as message passing graph neural networks without loss of generality for the approximation question."}},"verdict_id":"b26cf5a3-c12a-4dd6-90a1-3f941da48302"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:a281a6323011335accb87f7241de26396bf4848cc181ee5fe6897affa76d9ad7","target":"record","created_at":"2026-05-20T00:05:41Z","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":"d401ffe1572aa2b0b91d614c54be15dab4df12a69967d262d644a3b69113f001","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-07T16:52:38Z","title_canon_sha256":"400bd5230f4c4ac176391aeb223bfe9a66d02d7e527c151b9cddd0ace5081e39"},"schema_version":"1.0","source":{"id":"2602.07618","kind":"arxiv","version":6}},"canonical_sha256":"759eae04b8e7c493c2a4af03819fd9f8e792acd6b7cdd40d4aea29745088d97d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"759eae04b8e7c493c2a4af03819fd9f8e792acd6b7cdd40d4aea29745088d97d","first_computed_at":"2026-05-20T00:05:41.790668Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:05:41.790668Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"nBmBirZpD7oKtVA3+lJb3o5I1vIPpf0zcBPfp1j5zIGKf7+yFo3sMSJUxfAj/Jp41QPFaEsb3hLC+ZX/SsJ/CA==","signature_status":"signed_v1","signed_at":"2026-05-20T00:05:41.791117Z","signed_message":"canonical_sha256_bytes"},"source_id":"2602.07618","source_kind":"arxiv","source_version":6}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a281a6323011335accb87f7241de26396bf4848cc181ee5fe6897affa76d9ad7","sha256:d30b2e2f7e661c4b2164a8509e07b28ed21a653fc3c2ab5ace1fbc64e92bdd7d"],"state_sha256":"81a416c587e754facab57c9072517fb28fd10eb49981d0cffae3ef8a2b054eb7"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"gNtlRvLIUqt7bJIMHLs7JBh5EBEcol0qncv++FSkbhSH8GMRNmMMg++pcXtZGxavCCS5h5rXswvDsnMvzSclCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-27T18:42:42.150503Z","bundle_sha256":"a76967153f8d970c18bdbbce0cb95a1a3df0a7de077c28e79ec1b7072bf6d591"}}