{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IKST5AYPG6ZVUY7HU6RF5AVEOV","short_pith_number":"pith:IKST5AYP","schema_version":"1.0","canonical_sha256":"42a53e830f37b35a63e7a7a25e82a47562f27a733d763915aa6d6bd0935c01b1","source":{"kind":"arxiv","id":"2505.20892","version":1},"attestation_state":"computed","paper":{"title":"One-Time Soft Alignment Enables Resilient Learning without Weight Transport","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Jaehyuk Bae, Jeonghwan Cheon, Se-Bum Paik","submitted_at":"2025-05-27T08:34:12Z","abstract_excerpt":"Backpropagation is the cornerstone of deep learning, but its reliance on symmetric weight transport and global synchronization makes it computationally expensive and biologically implausible. Feedback alignment offers a promising alternative by approximating error gradients through fixed random feedback, thereby avoiding symmetric weight transport. However, this approach often struggles with poor learning performance and instability, especially in deep networks. Here, we show that a one-time soft alignment between forward and feedback weights at initialization enables deep networks to achieve "},"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":"2505.20892","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-27T08:34:12Z","cross_cats_sorted":[],"title_canon_sha256":"78972c04a9ea0f248f27d93d58892ddaee824a0214f74863ce04032b79fe0243","abstract_canon_sha256":"a5430c933820b4c0118b53b96ea2fe5592417f91e8d657135ea07ef5f88e2d92"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:10:20.568981Z","signature_b64":"PW0T0uXldbqGPuRbAas9IF7ko7xl5CVLF11FDl/cfiq8VcwJt4i531CWEMG0pkbXJfAjsBVbWXC6XJcd9mbKBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"42a53e830f37b35a63e7a7a25e82a47562f27a733d763915aa6d6bd0935c01b1","last_reissued_at":"2026-07-05T11:10:20.568430Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:10:20.568430Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"One-Time Soft Alignment Enables Resilient Learning without Weight Transport","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Jaehyuk Bae, Jeonghwan Cheon, Se-Bum Paik","submitted_at":"2025-05-27T08:34:12Z","abstract_excerpt":"Backpropagation is the cornerstone of deep learning, but its reliance on symmetric weight transport and global synchronization makes it computationally expensive and biologically implausible. Feedback alignment offers a promising alternative by approximating error gradients through fixed random feedback, thereby avoiding symmetric weight transport. However, this approach often struggles with poor learning performance and instability, especially in deep networks. Here, we show that a one-time soft alignment between forward and feedback weights at initialization enables deep networks to achieve "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.20892","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/2505.20892/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":"2505.20892","created_at":"2026-07-05T11:10:20.568488+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.20892v1","created_at":"2026-07-05T11:10:20.568488+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.20892","created_at":"2026-07-05T11:10:20.568488+00:00"},{"alias_kind":"pith_short_12","alias_value":"IKST5AYPG6ZV","created_at":"2026-07-05T11:10:20.568488+00:00"},{"alias_kind":"pith_short_16","alias_value":"IKST5AYPG6ZVUY7H","created_at":"2026-07-05T11:10:20.568488+00:00"},{"alias_kind":"pith_short_8","alias_value":"IKST5AYP","created_at":"2026-07-05T11:10:20.568488+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2508.01219","citing_title":"Eigen Neural Network: Unlocking Generalizable Vision with Eigenbasis","ref_index":10,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IKST5AYPG6ZVUY7HU6RF5AVEOV","json":"https://pith.science/pith/IKST5AYPG6ZVUY7HU6RF5AVEOV.json","graph_json":"https://pith.science/api/pith-number/IKST5AYPG6ZVUY7HU6RF5AVEOV/graph.json","events_json":"https://pith.science/api/pith-number/IKST5AYPG6ZVUY7HU6RF5AVEOV/events.json","paper":"https://pith.science/paper/IKST5AYP"},"agent_actions":{"view_html":"https://pith.science/pith/IKST5AYPG6ZVUY7HU6RF5AVEOV","download_json":"https://pith.science/pith/IKST5AYPG6ZVUY7HU6RF5AVEOV.json","view_paper":"https://pith.science/paper/IKST5AYP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.20892&json=true","fetch_graph":"https://pith.science/api/pith-number/IKST5AYPG6ZVUY7HU6RF5AVEOV/graph.json","fetch_events":"https://pith.science/api/pith-number/IKST5AYPG6ZVUY7HU6RF5AVEOV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IKST5AYPG6ZVUY7HU6RF5AVEOV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IKST5AYPG6ZVUY7HU6RF5AVEOV/action/storage_attestation","attest_author":"https://pith.science/pith/IKST5AYPG6ZVUY7HU6RF5AVEOV/action/author_attestation","sign_citation":"https://pith.science/pith/IKST5AYPG6ZVUY7HU6RF5AVEOV/action/citation_signature","submit_replication":"https://pith.science/pith/IKST5AYPG6ZVUY7HU6RF5AVEOV/action/replication_record"}},"created_at":"2026-07-05T11:10:20.568488+00:00","updated_at":"2026-07-05T11:10:20.568488+00:00"}