{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:4ZIPUEERTSICM35NQS3PB3XRGJ","short_pith_number":"pith:4ZIPUEER","schema_version":"1.0","canonical_sha256":"e650fa10919c90266fad84b6f0eef13264abac3e6621e5851c227395bdfc739f","source":{"kind":"arxiv","id":"2312.02364","version":3},"attestation_state":"computed","paper":{"title":"Class-Discriminative Attention Maps for Vision Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG","stat.ML"],"primary_cat":"cs.CV","authors_text":"Jakub Binda, Lennart Brocki, Neo Christopher Chung","submitted_at":"2023-12-04T21:46:21Z","abstract_excerpt":"Importance estimators are explainability methods that quantify feature importance for deep neural networks (DNN). In vision transformers (ViT), the self-attention mechanism naturally leads to attention maps, which are sometimes interpreted as importance scores that indicate which input features ViT models are focusing on. However, attention maps do not account for signals from downstream tasks. To generate explanations that are sensitive to downstream tasks, we have developed class-discriminative attention maps (CDAM), a gradient-based extension that estimates feature importance with respect 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":"2312.02364","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-12-04T21:46:21Z","cross_cats_sorted":["cs.AI","cs.LG","stat.ML"],"title_canon_sha256":"ea3de8323de9336133e891928afe1d67e0f720eeb1792a7061be9a3d46c4f59c","abstract_canon_sha256":"f3422664b510287e85562c47d1458eb22f4dcccbf9a8d22f50a7899b6f9a9989"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:25:36.971909Z","signature_b64":"9tKODLFg2e1oCO9mkla3dYdQLRkPfxNhC5FJBq6bmzx7EgN5AH/Fiw58GEif95uM68NY2eoVcNqYRpT7m4vNCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e650fa10919c90266fad84b6f0eef13264abac3e6621e5851c227395bdfc739f","last_reissued_at":"2026-07-05T09:25:36.971444Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:25:36.971444Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Class-Discriminative Attention Maps for Vision Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG","stat.ML"],"primary_cat":"cs.CV","authors_text":"Jakub Binda, Lennart Brocki, Neo Christopher Chung","submitted_at":"2023-12-04T21:46:21Z","abstract_excerpt":"Importance estimators are explainability methods that quantify feature importance for deep neural networks (DNN). In vision transformers (ViT), the self-attention mechanism naturally leads to attention maps, which are sometimes interpreted as importance scores that indicate which input features ViT models are focusing on. However, attention maps do not account for signals from downstream tasks. To generate explanations that are sensitive to downstream tasks, we have developed class-discriminative attention maps (CDAM), a gradient-based extension that estimates feature importance with respect t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.02364","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/2312.02364/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":"2312.02364","created_at":"2026-07-05T09:25:36.971501+00:00"},{"alias_kind":"arxiv_version","alias_value":"2312.02364v3","created_at":"2026-07-05T09:25:36.971501+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.02364","created_at":"2026-07-05T09:25:36.971501+00:00"},{"alias_kind":"pith_short_12","alias_value":"4ZIPUEERTSIC","created_at":"2026-07-05T09:25:36.971501+00:00"},{"alias_kind":"pith_short_16","alias_value":"4ZIPUEERTSICM35N","created_at":"2026-07-05T09:25:36.971501+00:00"},{"alias_kind":"pith_short_8","alias_value":"4ZIPUEER","created_at":"2026-07-05T09:25:36.971501+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.18094","citing_title":"Decision-Aware Attention Propagation for Vision Transformer Explainability","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4ZIPUEERTSICM35NQS3PB3XRGJ","json":"https://pith.science/pith/4ZIPUEERTSICM35NQS3PB3XRGJ.json","graph_json":"https://pith.science/api/pith-number/4ZIPUEERTSICM35NQS3PB3XRGJ/graph.json","events_json":"https://pith.science/api/pith-number/4ZIPUEERTSICM35NQS3PB3XRGJ/events.json","paper":"https://pith.science/paper/4ZIPUEER"},"agent_actions":{"view_html":"https://pith.science/pith/4ZIPUEERTSICM35NQS3PB3XRGJ","download_json":"https://pith.science/pith/4ZIPUEERTSICM35NQS3PB3XRGJ.json","view_paper":"https://pith.science/paper/4ZIPUEER","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2312.02364&json=true","fetch_graph":"https://pith.science/api/pith-number/4ZIPUEERTSICM35NQS3PB3XRGJ/graph.json","fetch_events":"https://pith.science/api/pith-number/4ZIPUEERTSICM35NQS3PB3XRGJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4ZIPUEERTSICM35NQS3PB3XRGJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4ZIPUEERTSICM35NQS3PB3XRGJ/action/storage_attestation","attest_author":"https://pith.science/pith/4ZIPUEERTSICM35NQS3PB3XRGJ/action/author_attestation","sign_citation":"https://pith.science/pith/4ZIPUEERTSICM35NQS3PB3XRGJ/action/citation_signature","submit_replication":"https://pith.science/pith/4ZIPUEERTSICM35NQS3PB3XRGJ/action/replication_record"}},"created_at":"2026-07-05T09:25:36.971501+00:00","updated_at":"2026-07-05T09:25:36.971501+00:00"}