{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:T6HXDXI5T6DCCNYEMDWBKLT24O","short_pith_number":"pith:T6HXDXI5","schema_version":"1.0","canonical_sha256":"9f8f71dd1d9f8621370460ec152e7ae38ebf59870be78a60f56e61b6503c9373","source":{"kind":"arxiv","id":"2604.15166","version":2},"attestation_state":"computed","paper":{"title":"Class Unlearning via Depth-Aware Removal of Forget-Specific Directions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"By projecting out forget-specific directions layer by layer with depth-aware scaling, a closed-form method achieves class unlearning closer to full retraining than prior approaches.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Arman Hatami, Ilya E. Monosov, Romina Aalishah","submitted_at":"2026-04-16T15:46:02Z","abstract_excerpt":"Machine unlearning aims to remove targeted knowledge from a trained model without the cost of retraining from scratch. In class unlearning, however, reducing accuracy on forget classes does not necessarily imply true forgetting: forgotten information can remain encoded in internal representations, and apparent forgetting may arise from classifier-head suppression rather than representational removal. We show that existing class-unlearning methods often exhibit weak or negative selectivity, preserve forget-class structure in deep representations, or rely heavily on final-layer bias shifts. We t"},"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":"2604.15166","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-04-16T15:46:02Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"9599830383a0f2641a0b445a59d5275738d96d4977a5d3214365f5153c26e668","abstract_canon_sha256":"c0ad3f258483854b11b7e8c6a5cfb15767e9be967efcd7e68a2d2e6a8ac57a69"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T02:05:43.204758Z","signature_b64":"hOoy4w14JwQN+29joNgmkVS+HyUbj/kgKF/5ITuVFLIqOwfYD/DbetMgs1SzUzF2vot0sWFf2dboFJNiTsTADg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9f8f71dd1d9f8621370460ec152e7ae38ebf59870be78a60f56e61b6503c9373","last_reissued_at":"2026-05-20T02:05:43.203711Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T02:05:43.203711Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Class Unlearning via Depth-Aware Removal of Forget-Specific Directions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"By projecting out forget-specific directions layer by layer with depth-aware scaling, a closed-form method achieves class unlearning closer to full retraining than prior approaches.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Arman Hatami, Ilya E. Monosov, Romina Aalishah","submitted_at":"2026-04-16T15:46:02Z","abstract_excerpt":"Machine unlearning aims to remove targeted knowledge from a trained model without the cost of retraining from scratch. In class unlearning, however, reducing accuracy on forget classes does not necessarily imply true forgetting: forgotten information can remain encoded in internal representations, and apparent forgetting may arise from classifier-head suppression rather than representational removal. We show that existing class-unlearning methods often exhibit weak or negative selectivity, preserve forget-class structure in deep representations, or rely heavily on final-layer bias shifts. We t"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Across MNIST, CIFAR-10, CIFAR-100, and Tiny ImageNet, and across convolutional and transformer architectures, DAMP more closely resembles the retraining gold standard than some of the prior methods, improving selective forgetting while better preserving retain-class performance and reducing residual forget-class structure in deep layers.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That forget directions extracted as residuals relative to retain-class prototypes at each layer accurately isolate the targeted knowledge, and that the parameter-free depth-aware scaling derived from probe separability will preserve utility without introducing new failure modes on retain classes.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"DAMP performs one-shot class unlearning by extracting and projecting out forget-specific residual directions at each network depth using class prototypes and a separability-derived scaling rule.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"By projecting out forget-specific directions layer by layer with depth-aware scaling, a closed-form method achieves class unlearning closer to full retraining than prior approaches.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"b40c76d111c9bac4a0b4b8059e503eca22e0e488a746da4a466244dc9022aee2"},"source":{"id":"2604.15166","kind":"arxiv","version":2},"verdict":{"id":"ce474362-0a7f-4108-bf0c-ce88d88df969","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T11:42:48.804913Z","strongest_claim":"Across MNIST, CIFAR-10, CIFAR-100, and Tiny ImageNet, and across convolutional and transformer architectures, DAMP more closely resembles the retraining gold standard than some of the prior methods, improving selective forgetting while better preserving retain-class performance and reducing residual forget-class structure in deep layers.","one_line_summary":"DAMP performs one-shot class unlearning by extracting and projecting out forget-specific residual directions at each network depth using class prototypes and a separability-derived scaling rule.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That forget directions extracted as residuals relative to retain-class prototypes at each layer accurately isolate the targeted knowledge, and that the parameter-free depth-aware scaling derived from probe separability will preserve utility without introducing new failure modes on retain classes.","pith_extraction_headline":"By projecting out forget-specific directions layer by layer with depth-aware scaling, a closed-form method achieves class unlearning closer to full retraining than prior approaches."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.15166/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":"2604.15166","created_at":"2026-05-20T02:05:43.203842+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.15166v2","created_at":"2026-05-20T02:05:43.203842+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.15166","created_at":"2026-05-20T02:05:43.203842+00:00"},{"alias_kind":"pith_short_12","alias_value":"T6HXDXI5T6DC","created_at":"2026-05-20T02:05:43.203842+00:00"},{"alias_kind":"pith_short_16","alias_value":"T6HXDXI5T6DCCNYE","created_at":"2026-05-20T02:05:43.203842+00:00"},{"alias_kind":"pith_short_8","alias_value":"T6HXDXI5","created_at":"2026-05-20T02:05:43.203842+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.08730","citing_title":"Classification-Head Bias in Class-Level Machine Unlearning: Diagnosis, Mitigation, and Evaluation","ref_index":4,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/T6HXDXI5T6DCCNYEMDWBKLT24O","json":"https://pith.science/pith/T6HXDXI5T6DCCNYEMDWBKLT24O.json","graph_json":"https://pith.science/api/pith-number/T6HXDXI5T6DCCNYEMDWBKLT24O/graph.json","events_json":"https://pith.science/api/pith-number/T6HXDXI5T6DCCNYEMDWBKLT24O/events.json","paper":"https://pith.science/paper/T6HXDXI5"},"agent_actions":{"view_html":"https://pith.science/pith/T6HXDXI5T6DCCNYEMDWBKLT24O","download_json":"https://pith.science/pith/T6HXDXI5T6DCCNYEMDWBKLT24O.json","view_paper":"https://pith.science/paper/T6HXDXI5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.15166&json=true","fetch_graph":"https://pith.science/api/pith-number/T6HXDXI5T6DCCNYEMDWBKLT24O/graph.json","fetch_events":"https://pith.science/api/pith-number/T6HXDXI5T6DCCNYEMDWBKLT24O/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/T6HXDXI5T6DCCNYEMDWBKLT24O/action/timestamp_anchor","attest_storage":"https://pith.science/pith/T6HXDXI5T6DCCNYEMDWBKLT24O/action/storage_attestation","attest_author":"https://pith.science/pith/T6HXDXI5T6DCCNYEMDWBKLT24O/action/author_attestation","sign_citation":"https://pith.science/pith/T6HXDXI5T6DCCNYEMDWBKLT24O/action/citation_signature","submit_replication":"https://pith.science/pith/T6HXDXI5T6DCCNYEMDWBKLT24O/action/replication_record"}},"created_at":"2026-05-20T02:05:43.203842+00:00","updated_at":"2026-05-20T02:05:43.203842+00:00"}