{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:AF5ZWFNRIRNWIDUE3BXZIJIIZ6","short_pith_number":"pith:AF5ZWFNR","schema_version":"1.0","canonical_sha256":"017b9b15b1445b640e84d86f942508cf8849dadcfbe758507d865785ff940bb3","source":{"kind":"arxiv","id":"2505.07450","version":3},"attestation_state":"computed","paper":{"title":"Prototype Augmented Hypernetworks for Continual Learning","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Albin Soutiff, Andrey Barsky, Cecilia Curreli, Daniel Cremers, Daniel Vidal, Maria Pilligua, Neil De La Fuente","submitted_at":"2025-05-12T11:25:54Z","abstract_excerpt":"Continual learning (CL) aims to learn a sequence of tasks without forgetting prior knowledge, but gradient updates for a new task often overwrite the weights learned earlier, causing catastrophic forgetting (CF). We propose Prototype-Augmented Hypernetworks (PAH), a framework where a single hypernetwork, conditioned on learnable task prototypes, dynamically generates task-specific classifier heads on demand. To mitigate forgetting, PAH combines cross-entropy with dual distillation losses, one to align logits and another to align prototypes, ensuring stable feature representations across tasks."},"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":"2505.07450","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-12T11:25:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d8093e1b93c5092dd5e6c0ceaca4d44c98daae9b06809791fa1e5260cac1129b","abstract_canon_sha256":"966748b50652802a301dc64d8feea440b4327bc89b5c5b9780addeafd7622e53"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:04:00.649752Z","signature_b64":"gnqE5eLHgZ/c4XqVZnEHDhimu6qumGLNp9JAI7xhZDx4ejbv/IaiwF+8DIGGFxaOsrTK8WNTgo0CrlL6TVWEBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"017b9b15b1445b640e84d86f942508cf8849dadcfbe758507d865785ff940bb3","last_reissued_at":"2026-07-05T11:04:00.649245Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:04:00.649245Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Prototype Augmented Hypernetworks for Continual Learning","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Albin Soutiff, Andrey Barsky, Cecilia Curreli, Daniel Cremers, Daniel Vidal, Maria Pilligua, Neil De La Fuente","submitted_at":"2025-05-12T11:25:54Z","abstract_excerpt":"Continual learning (CL) aims to learn a sequence of tasks without forgetting prior knowledge, but gradient updates for a new task often overwrite the weights learned earlier, causing catastrophic forgetting (CF). We propose Prototype-Augmented Hypernetworks (PAH), a framework where a single hypernetwork, conditioned on learnable task prototypes, dynamically generates task-specific classifier heads on demand. To mitigate forgetting, PAH combines cross-entropy with dual distillation losses, one to align logits and another to align prototypes, ensuring stable feature representations across tasks."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.07450","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2505.07450/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":"2505.07450","created_at":"2026-07-05T11:04:00.649305+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.07450v3","created_at":"2026-07-05T11:04:00.649305+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.07450","created_at":"2026-07-05T11:04:00.649305+00:00"},{"alias_kind":"pith_short_12","alias_value":"AF5ZWFNRIRNW","created_at":"2026-07-05T11:04:00.649305+00:00"},{"alias_kind":"pith_short_16","alias_value":"AF5ZWFNRIRNWIDUE","created_at":"2026-07-05T11:04:00.649305+00:00"},{"alias_kind":"pith_short_8","alias_value":"AF5ZWFNR","created_at":"2026-07-05T11:04:00.649305+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.15626","citing_title":"Aligning Generalisation Between Humans and Machines","ref_index":51,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AF5ZWFNRIRNWIDUE3BXZIJIIZ6","json":"https://pith.science/pith/AF5ZWFNRIRNWIDUE3BXZIJIIZ6.json","graph_json":"https://pith.science/api/pith-number/AF5ZWFNRIRNWIDUE3BXZIJIIZ6/graph.json","events_json":"https://pith.science/api/pith-number/AF5ZWFNRIRNWIDUE3BXZIJIIZ6/events.json","paper":"https://pith.science/paper/AF5ZWFNR"},"agent_actions":{"view_html":"https://pith.science/pith/AF5ZWFNRIRNWIDUE3BXZIJIIZ6","download_json":"https://pith.science/pith/AF5ZWFNRIRNWIDUE3BXZIJIIZ6.json","view_paper":"https://pith.science/paper/AF5ZWFNR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.07450&json=true","fetch_graph":"https://pith.science/api/pith-number/AF5ZWFNRIRNWIDUE3BXZIJIIZ6/graph.json","fetch_events":"https://pith.science/api/pith-number/AF5ZWFNRIRNWIDUE3BXZIJIIZ6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AF5ZWFNRIRNWIDUE3BXZIJIIZ6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AF5ZWFNRIRNWIDUE3BXZIJIIZ6/action/storage_attestation","attest_author":"https://pith.science/pith/AF5ZWFNRIRNWIDUE3BXZIJIIZ6/action/author_attestation","sign_citation":"https://pith.science/pith/AF5ZWFNRIRNWIDUE3BXZIJIIZ6/action/citation_signature","submit_replication":"https://pith.science/pith/AF5ZWFNRIRNWIDUE3BXZIJIIZ6/action/replication_record"}},"created_at":"2026-07-05T11:04:00.649305+00:00","updated_at":"2026-07-05T11:04:00.649305+00:00"}