{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2016:C3PIOFLSHOAAY6ABVCYOHC573A","short_pith_number":"pith:C3PIOFLS","schema_version":"1.0","canonical_sha256":"16de8715723b800c7801a8b0e38bbfd83b27f16491993f80538430d586227e20","source":{"kind":"arxiv","id":"1603.05027","version":3},"attestation_state":"computed","paper":{"title":"Identity Mappings in Deep Residual Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Jian Sun, Kaiming He, Shaoqing Ren, Xiangyu Zhang","submitted_at":"2016-03-16T10:53:56Z","abstract_excerpt":"Deep residual networks have emerged as a family of extremely deep architectures showing compelling accuracy and nice convergence behaviors. In this paper, we analyze the propagation formulations behind the residual building blocks, which suggest that the forward and backward signals can be directly propagated from one block to any other block, when using identity mappings as the skip connections and after-addition activation. A series of ablation experiments support the importance of these identity mappings. This motivates us to propose a new residual unit, which makes training easier and impr"},"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":"1603.05027","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2016-03-16T10:53:56Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"abebf9356333105ad4eb147b0d3c237242bc7e1aaae68c18fa0bc814a925eb2f","abstract_canon_sha256":"80ce869d30442a05b9686a036121a7a793ae79e29ce0005a94bd4034e15b7d40"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T01:10:37.841350Z","signature_b64":"ChqYHarZDdU7ToUdZs7Jj/PJb754ZJtYsgJjtfcQ4ivOa9B5JX51YViSwK5iE3MEZ/9hoq5TV+qoWPy+ouXmCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"16de8715723b800c7801a8b0e38bbfd83b27f16491993f80538430d586227e20","last_reissued_at":"2026-05-18T01:10:37.840770Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T01:10:37.840770Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Identity Mappings in Deep Residual Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Jian Sun, Kaiming He, Shaoqing Ren, Xiangyu Zhang","submitted_at":"2016-03-16T10:53:56Z","abstract_excerpt":"Deep residual networks have emerged as a family of extremely deep architectures showing compelling accuracy and nice convergence behaviors. In this paper, we analyze the propagation formulations behind the residual building blocks, which suggest that the forward and backward signals can be directly propagated from one block to any other block, when using identity mappings as the skip connections and after-addition activation. A series of ablation experiments support the importance of these identity mappings. This motivates us to propose a new residual unit, which makes training easier and impr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1603.05027","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":"1603.05027","created_at":"2026-05-18T01:10:37.840864+00:00"},{"alias_kind":"arxiv_version","alias_value":"1603.05027v3","created_at":"2026-05-18T01:10:37.840864+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1603.05027","created_at":"2026-05-18T01:10:37.840864+00:00"},{"alias_kind":"pith_short_12","alias_value":"C3PIOFLSHOAA","created_at":"2026-05-18T12:30:09.641336+00:00"},{"alias_kind":"pith_short_16","alias_value":"C3PIOFLSHOAAY6AB","created_at":"2026-05-18T12:30:09.641336+00:00"},{"alias_kind":"pith_short_8","alias_value":"C3PIOFLS","created_at":"2026-05-18T12:30:09.641336+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2510.01712","citing_title":"ActiNet: An Open-Source Tool for Activity Intensity Classification of Wrist-Worn Accelerometry Using Self-Supervised Deep Learning","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"1605.07146","citing_title":"Wide Residual Networks","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"1604.06174","citing_title":"Training Deep Nets with Sublinear Memory Cost","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08300","citing_title":"mHC-SSM: Manifold-Constrained Hyper-Connections for State Space Language Models with Stream-Specialized Adapters","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"1605.08803","citing_title":"Density estimation using Real NVP","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"1904.10509","citing_title":"Generating Long Sequences with Sparse Transformers","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"1608.03983","citing_title":"SGDR: Stochastic Gradient Descent with Warm Restarts","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/C3PIOFLSHOAAY6ABVCYOHC573A","json":"https://pith.science/pith/C3PIOFLSHOAAY6ABVCYOHC573A.json","graph_json":"https://pith.science/api/pith-number/C3PIOFLSHOAAY6ABVCYOHC573A/graph.json","events_json":"https://pith.science/api/pith-number/C3PIOFLSHOAAY6ABVCYOHC573A/events.json","paper":"https://pith.science/paper/C3PIOFLS"},"agent_actions":{"view_html":"https://pith.science/pith/C3PIOFLSHOAAY6ABVCYOHC573A","download_json":"https://pith.science/pith/C3PIOFLSHOAAY6ABVCYOHC573A.json","view_paper":"https://pith.science/paper/C3PIOFLS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1603.05027&json=true","fetch_graph":"https://pith.science/api/pith-number/C3PIOFLSHOAAY6ABVCYOHC573A/graph.json","fetch_events":"https://pith.science/api/pith-number/C3PIOFLSHOAAY6ABVCYOHC573A/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/C3PIOFLSHOAAY6ABVCYOHC573A/action/timestamp_anchor","attest_storage":"https://pith.science/pith/C3PIOFLSHOAAY6ABVCYOHC573A/action/storage_attestation","attest_author":"https://pith.science/pith/C3PIOFLSHOAAY6ABVCYOHC573A/action/author_attestation","sign_citation":"https://pith.science/pith/C3PIOFLSHOAAY6ABVCYOHC573A/action/citation_signature","submit_replication":"https://pith.science/pith/C3PIOFLSHOAAY6ABVCYOHC573A/action/replication_record"}},"created_at":"2026-05-18T01:10:37.840864+00:00","updated_at":"2026-05-18T01:10:37.840864+00:00"}