{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:QNO3HCCFSAE3UWJKZKRDL3CTPN","short_pith_number":"pith:QNO3HCCF","schema_version":"1.0","canonical_sha256":"835db388459009ba592acaa235ec537b5a897832d9923d8fef8f4b622b78e06f","source":{"kind":"arxiv","id":"1805.07477","version":5},"attestation_state":"computed","paper":{"title":"Norm-Preservation: Why Residual Networks Can Become Extremely Deep?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alireza Zaeemzadeh, Mubarak Shah, Nazanin Rahnavard","submitted_at":"2018-05-18T23:37:17Z","abstract_excerpt":"Augmenting neural networks with skip connections, as introduced in the so-called ResNet architecture, surprised the community by enabling the training of networks of more than 1,000 layers with significant performance gains. This paper deciphers ResNet by analyzing the effect of skip connections, and puts forward new theoretical results on the advantages of identity skip connections in neural networks. We prove that the skip connections in the residual blocks facilitate preserving the norm of the gradient, and lead to stable back-propagation, which is desirable from optimization perspective. W"},"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":"1805.07477","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2018-05-18T23:37:17Z","cross_cats_sorted":[],"title_canon_sha256":"82467d48539da73ef73bbfff2709a364c708051c5185fe5b6585ef0b93bf6c01","abstract_canon_sha256":"4c17df24d677a10e68f2997f283daded0cca37a5d441d2003a59d2945df44862"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:57:32.348456Z","signature_b64":"dg3RkzOda+jx95H5e4162bqlpfFeHAgD5Z55r6JWxY7xpmO/4mpC05OOn8LFZ/HndvM0dQ1pIl0qbqseasDfCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"835db388459009ba592acaa235ec537b5a897832d9923d8fef8f4b622b78e06f","last_reissued_at":"2026-07-05T00:57:32.347992Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:57:32.347992Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Norm-Preservation: Why Residual Networks Can Become Extremely Deep?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alireza Zaeemzadeh, Mubarak Shah, Nazanin Rahnavard","submitted_at":"2018-05-18T23:37:17Z","abstract_excerpt":"Augmenting neural networks with skip connections, as introduced in the so-called ResNet architecture, surprised the community by enabling the training of networks of more than 1,000 layers with significant performance gains. This paper deciphers ResNet by analyzing the effect of skip connections, and puts forward new theoretical results on the advantages of identity skip connections in neural networks. We prove that the skip connections in the residual blocks facilitate preserving the norm of the gradient, and lead to stable back-propagation, which is desirable from optimization perspective. W"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1805.07477","kind":"arxiv","version":5},"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/1805.07477/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":"1805.07477","created_at":"2026-07-05T00:57:32.348053+00:00"},{"alias_kind":"arxiv_version","alias_value":"1805.07477v5","created_at":"2026-07-05T00:57:32.348053+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1805.07477","created_at":"2026-07-05T00:57:32.348053+00:00"},{"alias_kind":"pith_short_12","alias_value":"QNO3HCCFSAE3","created_at":"2026-07-05T00:57:32.348053+00:00"},{"alias_kind":"pith_short_16","alias_value":"QNO3HCCFSAE3UWJK","created_at":"2026-07-05T00:57:32.348053+00:00"},{"alias_kind":"pith_short_8","alias_value":"QNO3HCCF","created_at":"2026-07-05T00:57:32.348053+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"1908.11365","citing_title":"Improving Deep Transformer with Depth-Scaled Initialization and Merged Attention","ref_index":39,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QNO3HCCFSAE3UWJKZKRDL3CTPN","json":"https://pith.science/pith/QNO3HCCFSAE3UWJKZKRDL3CTPN.json","graph_json":"https://pith.science/api/pith-number/QNO3HCCFSAE3UWJKZKRDL3CTPN/graph.json","events_json":"https://pith.science/api/pith-number/QNO3HCCFSAE3UWJKZKRDL3CTPN/events.json","paper":"https://pith.science/paper/QNO3HCCF"},"agent_actions":{"view_html":"https://pith.science/pith/QNO3HCCFSAE3UWJKZKRDL3CTPN","download_json":"https://pith.science/pith/QNO3HCCFSAE3UWJKZKRDL3CTPN.json","view_paper":"https://pith.science/paper/QNO3HCCF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1805.07477&json=true","fetch_graph":"https://pith.science/api/pith-number/QNO3HCCFSAE3UWJKZKRDL3CTPN/graph.json","fetch_events":"https://pith.science/api/pith-number/QNO3HCCFSAE3UWJKZKRDL3CTPN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QNO3HCCFSAE3UWJKZKRDL3CTPN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QNO3HCCFSAE3UWJKZKRDL3CTPN/action/storage_attestation","attest_author":"https://pith.science/pith/QNO3HCCFSAE3UWJKZKRDL3CTPN/action/author_attestation","sign_citation":"https://pith.science/pith/QNO3HCCFSAE3UWJKZKRDL3CTPN/action/citation_signature","submit_replication":"https://pith.science/pith/QNO3HCCFSAE3UWJKZKRDL3CTPN/action/replication_record"}},"created_at":"2026-07-05T00:57:32.348053+00:00","updated_at":"2026-07-05T00:57:32.348053+00:00"}