{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2014:PO25V4RTJ35BQTTU4XIBLR3V66","short_pith_number":"pith:PO25V4RT","schema_version":"1.0","canonical_sha256":"7bb5daf2334efa184e74e5d015c775f7b9c39b0660d502170391429b7a6131f6","source":{"kind":"arxiv","id":"1412.6558","version":3},"attestation_state":"computed","paper":{"title":"Random Walk Initialization for Training Very Deep Feedforward Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ML"],"primary_cat":"cs.NE","authors_text":"David Sussillo, L.F. Abbott","submitted_at":"2014-12-19T23:24:53Z","abstract_excerpt":"Training very deep networks is an important open problem in machine learning. One of many difficulties is that the norm of the back-propagated error gradient can grow or decay exponentially. Here we show that training very deep feed-forward networks (FFNs) is not as difficult as previously thought. Unlike when back-propagation is applied to a recurrent network, application to an FFN amounts to multiplying the error gradient by a different random matrix at each layer. We show that the successive application of correctly scaled random matrices to an initial vector results in a random walk of the"},"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":"1412.6558","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.NE","submitted_at":"2014-12-19T23:24:53Z","cross_cats_sorted":["cs.LG","stat.ML"],"title_canon_sha256":"65bb8465f2e987a7e38c32e5bee478c5e6583d7c5a1312becb439e74897ac429","abstract_canon_sha256":"5c723eee5207c159466aca8aab68ca040b495e2bd66ba63f74ad4b2622e09c30"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:25:55.675505Z","signature_b64":"mVY98kj6vValW9qjy2RKZFaIKuj8Y3IAiLrVyz4OyLXx9awDGYA8K2xKj+s/Nkgu+uc8QxPU8ekE78p6Z7jOCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7bb5daf2334efa184e74e5d015c775f7b9c39b0660d502170391429b7a6131f6","last_reissued_at":"2026-05-18T02:25:55.675110Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:25:55.675110Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Random Walk Initialization for Training Very Deep Feedforward Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ML"],"primary_cat":"cs.NE","authors_text":"David Sussillo, L.F. Abbott","submitted_at":"2014-12-19T23:24:53Z","abstract_excerpt":"Training very deep networks is an important open problem in machine learning. One of many difficulties is that the norm of the back-propagated error gradient can grow or decay exponentially. Here we show that training very deep feed-forward networks (FFNs) is not as difficult as previously thought. Unlike when back-propagation is applied to a recurrent network, application to an FFN amounts to multiplying the error gradient by a different random matrix at each layer. We show that the successive application of correctly scaled random matrices to an initial vector results in a random walk of the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1412.6558","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"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":"1412.6558","created_at":"2026-05-18T02:25:55.675169+00:00"},{"alias_kind":"arxiv_version","alias_value":"1412.6558v3","created_at":"2026-05-18T02:25:55.675169+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1412.6558","created_at":"2026-05-18T02:25:55.675169+00:00"},{"alias_kind":"pith_short_12","alias_value":"PO25V4RTJ35B","created_at":"2026-05-18T12:28:43.426989+00:00"},{"alias_kind":"pith_short_16","alias_value":"PO25V4RTJ35BQTTU","created_at":"2026-05-18T12:28:43.426989+00:00"},{"alias_kind":"pith_short_8","alias_value":"PO25V4RT","created_at":"2026-05-18T12:28:43.426989+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PO25V4RTJ35BQTTU4XIBLR3V66","json":"https://pith.science/pith/PO25V4RTJ35BQTTU4XIBLR3V66.json","graph_json":"https://pith.science/api/pith-number/PO25V4RTJ35BQTTU4XIBLR3V66/graph.json","events_json":"https://pith.science/api/pith-number/PO25V4RTJ35BQTTU4XIBLR3V66/events.json","paper":"https://pith.science/paper/PO25V4RT"},"agent_actions":{"view_html":"https://pith.science/pith/PO25V4RTJ35BQTTU4XIBLR3V66","download_json":"https://pith.science/pith/PO25V4RTJ35BQTTU4XIBLR3V66.json","view_paper":"https://pith.science/paper/PO25V4RT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1412.6558&json=true","fetch_graph":"https://pith.science/api/pith-number/PO25V4RTJ35BQTTU4XIBLR3V66/graph.json","fetch_events":"https://pith.science/api/pith-number/PO25V4RTJ35BQTTU4XIBLR3V66/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PO25V4RTJ35BQTTU4XIBLR3V66/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PO25V4RTJ35BQTTU4XIBLR3V66/action/storage_attestation","attest_author":"https://pith.science/pith/PO25V4RTJ35BQTTU4XIBLR3V66/action/author_attestation","sign_citation":"https://pith.science/pith/PO25V4RTJ35BQTTU4XIBLR3V66/action/citation_signature","submit_replication":"https://pith.science/pith/PO25V4RTJ35BQTTU4XIBLR3V66/action/replication_record"}},"created_at":"2026-05-18T02:25:55.675169+00:00","updated_at":"2026-05-18T02:25:55.675169+00:00"}