{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:6SXCP6W3A2MSEZSLOPRUZJ7432","short_pith_number":"pith:6SXCP6W3","schema_version":"1.0","canonical_sha256":"f4ae27fadb069922664b73e34ca7fcdea2a1ea5c31303537dd703cc54e6b474b","source":{"kind":"arxiv","id":"1908.10030","version":1},"attestation_state":"computed","paper":{"title":"Finite size corrections for neural network Gaussian processes","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Joseph M. Antognini","submitted_at":"2019-08-27T04:49:23Z","abstract_excerpt":"There has been a recent surge of interest in modeling neural networks (NNs) as Gaussian processes. In the limit of a NN of infinite width the NN becomes equivalent to a Gaussian process. Here we demonstrate that for an ensemble of large, finite, fully connected networks with a single hidden layer the distribution of outputs at initialization is well described by a Gaussian perturbed by the fourth Hermite polynomial for weights drawn from a symmetric distribution. We show that the scale of the perturbation is inversely proportional to the number of units in the NN and that higher order terms de"},"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":"1908.10030","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-08-27T04:49:23Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"3d0895c346029a3e4e068aca18881e52e6c7157966ca15d0f15cd81a7ea3c0e3","abstract_canon_sha256":"c144f6375845aed2fb0cd44932780218f439ce3023c67f99c204bfe5ca393a3b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-04T23:59:53.112775Z","signature_b64":"DJ1Hb53PTsj6Z/zUZcJznqVS9VeAzUPz0ijWlFzsNZ0L7MsCaC0cxzNhLDM90fUzkEBMi2LvokDHpZk1wi4qCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f4ae27fadb069922664b73e34ca7fcdea2a1ea5c31303537dd703cc54e6b474b","last_reissued_at":"2026-07-04T23:59:53.112428Z","signature_status":"signed_v1","first_computed_at":"2026-07-04T23:59:53.112428Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Finite size corrections for neural network Gaussian processes","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Joseph M. Antognini","submitted_at":"2019-08-27T04:49:23Z","abstract_excerpt":"There has been a recent surge of interest in modeling neural networks (NNs) as Gaussian processes. In the limit of a NN of infinite width the NN becomes equivalent to a Gaussian process. Here we demonstrate that for an ensemble of large, finite, fully connected networks with a single hidden layer the distribution of outputs at initialization is well described by a Gaussian perturbed by the fourth Hermite polynomial for weights drawn from a symmetric distribution. We show that the scale of the perturbation is inversely proportional to the number of units in the NN and that higher order terms de"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1908.10030","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/1908.10030/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":"1908.10030","created_at":"2026-07-04T23:59:53.112480+00:00"},{"alias_kind":"arxiv_version","alias_value":"1908.10030v1","created_at":"2026-07-04T23:59:53.112480+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.10030","created_at":"2026-07-04T23:59:53.112480+00:00"},{"alias_kind":"pith_short_12","alias_value":"6SXCP6W3A2MS","created_at":"2026-07-04T23:59:53.112480+00:00"},{"alias_kind":"pith_short_16","alias_value":"6SXCP6W3A2MSEZSL","created_at":"2026-07-04T23:59:53.112480+00:00"},{"alias_kind":"pith_short_8","alias_value":"6SXCP6W3","created_at":"2026-07-04T23:59:53.112480+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.24072","citing_title":"Optimal Non-Asymptotic Edgeworth Expansions for Multivariate Neural Network Outputs","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6SXCP6W3A2MSEZSLOPRUZJ7432","json":"https://pith.science/pith/6SXCP6W3A2MSEZSLOPRUZJ7432.json","graph_json":"https://pith.science/api/pith-number/6SXCP6W3A2MSEZSLOPRUZJ7432/graph.json","events_json":"https://pith.science/api/pith-number/6SXCP6W3A2MSEZSLOPRUZJ7432/events.json","paper":"https://pith.science/paper/6SXCP6W3"},"agent_actions":{"view_html":"https://pith.science/pith/6SXCP6W3A2MSEZSLOPRUZJ7432","download_json":"https://pith.science/pith/6SXCP6W3A2MSEZSLOPRUZJ7432.json","view_paper":"https://pith.science/paper/6SXCP6W3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1908.10030&json=true","fetch_graph":"https://pith.science/api/pith-number/6SXCP6W3A2MSEZSLOPRUZJ7432/graph.json","fetch_events":"https://pith.science/api/pith-number/6SXCP6W3A2MSEZSLOPRUZJ7432/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6SXCP6W3A2MSEZSLOPRUZJ7432/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6SXCP6W3A2MSEZSLOPRUZJ7432/action/storage_attestation","attest_author":"https://pith.science/pith/6SXCP6W3A2MSEZSLOPRUZJ7432/action/author_attestation","sign_citation":"https://pith.science/pith/6SXCP6W3A2MSEZSLOPRUZJ7432/action/citation_signature","submit_replication":"https://pith.science/pith/6SXCP6W3A2MSEZSLOPRUZJ7432/action/replication_record"}},"created_at":"2026-07-04T23:59:53.112480+00:00","updated_at":"2026-07-04T23:59:53.112480+00:00"}