{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:2BEG2GYRBMPN7XGPY4OJNUJIFW","short_pith_number":"pith:2BEG2GYR","schema_version":"1.0","canonical_sha256":"d0486d1b110b1edfdccfc71c96d1282db7b615d189b47552e03f6eb9e71fefb7","source":{"kind":"arxiv","id":"2302.08508","version":2},"attestation_state":"computed","paper":{"title":"Sanity checks and improvements for patch visualisation in prototype-based image classification","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Georges Qu\\'enot (MRIM), Marie-Christine Rousset (SLIDE), MRIM), Romain Xu-Darme (LSL, Zakaria Chihani (LSL)","submitted_at":"2023-01-20T15:13:04Z","abstract_excerpt":"In this work, we perform an in-depth analysis of the visualisation methods implemented in two popular self-explaining models for visual classification based on prototypes - ProtoPNet and ProtoTree. Using two fine-grained datasets (CUB-200-2011 and Stanford Cars), we first show that such methods do not correctly identify the regions of interest inside of the images, and therefore do not reflect the model behaviour. Secondly, using a deletion metric, we demonstrate quantitatively that saliency methods such as Smoothgrads or PRP provide more faithful image patches. We also propose a new relevance"},"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":"2302.08508","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-01-20T15:13:04Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"7a7c741e4e9b673cac34282225a2dfcb56bc73d92e53ab1fac7e15685c4e2aa7","abstract_canon_sha256":"4678d9cef1174b30f0553f04fefa5c7db8a29929f6d9866823e97ece6807948f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:09:48.510831Z","signature_b64":"iYknFnNeQH98fnHWgFhS1kgSyKa+5Xpg1EceIYaoDdXko2SdfY4soCKZS8rOGm8RUDPdjQiD5+WPmTdKLo+9Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d0486d1b110b1edfdccfc71c96d1282db7b615d189b47552e03f6eb9e71fefb7","last_reissued_at":"2026-07-05T06:09:48.510349Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:09:48.510349Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sanity checks and improvements for patch visualisation in prototype-based image classification","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Georges Qu\\'enot (MRIM), Marie-Christine Rousset (SLIDE), MRIM), Romain Xu-Darme (LSL, Zakaria Chihani (LSL)","submitted_at":"2023-01-20T15:13:04Z","abstract_excerpt":"In this work, we perform an in-depth analysis of the visualisation methods implemented in two popular self-explaining models for visual classification based on prototypes - ProtoPNet and ProtoTree. Using two fine-grained datasets (CUB-200-2011 and Stanford Cars), we first show that such methods do not correctly identify the regions of interest inside of the images, and therefore do not reflect the model behaviour. Secondly, using a deletion metric, we demonstrate quantitatively that saliency methods such as Smoothgrads or PRP provide more faithful image patches. We also propose a new relevance"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.08508","kind":"arxiv","version":2},"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/2302.08508/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":"2302.08508","created_at":"2026-07-05T06:09:48.510406+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.08508v2","created_at":"2026-07-05T06:09:48.510406+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.08508","created_at":"2026-07-05T06:09:48.510406+00:00"},{"alias_kind":"pith_short_12","alias_value":"2BEG2GYRBMPN","created_at":"2026-07-05T06:09:48.510406+00:00"},{"alias_kind":"pith_short_16","alias_value":"2BEG2GYRBMPN7XGP","created_at":"2026-07-05T06:09:48.510406+00:00"},{"alias_kind":"pith_short_8","alias_value":"2BEG2GYR","created_at":"2026-07-05T06:09:48.510406+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2BEG2GYRBMPN7XGPY4OJNUJIFW","json":"https://pith.science/pith/2BEG2GYRBMPN7XGPY4OJNUJIFW.json","graph_json":"https://pith.science/api/pith-number/2BEG2GYRBMPN7XGPY4OJNUJIFW/graph.json","events_json":"https://pith.science/api/pith-number/2BEG2GYRBMPN7XGPY4OJNUJIFW/events.json","paper":"https://pith.science/paper/2BEG2GYR"},"agent_actions":{"view_html":"https://pith.science/pith/2BEG2GYRBMPN7XGPY4OJNUJIFW","download_json":"https://pith.science/pith/2BEG2GYRBMPN7XGPY4OJNUJIFW.json","view_paper":"https://pith.science/paper/2BEG2GYR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.08508&json=true","fetch_graph":"https://pith.science/api/pith-number/2BEG2GYRBMPN7XGPY4OJNUJIFW/graph.json","fetch_events":"https://pith.science/api/pith-number/2BEG2GYRBMPN7XGPY4OJNUJIFW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2BEG2GYRBMPN7XGPY4OJNUJIFW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2BEG2GYRBMPN7XGPY4OJNUJIFW/action/storage_attestation","attest_author":"https://pith.science/pith/2BEG2GYRBMPN7XGPY4OJNUJIFW/action/author_attestation","sign_citation":"https://pith.science/pith/2BEG2GYRBMPN7XGPY4OJNUJIFW/action/citation_signature","submit_replication":"https://pith.science/pith/2BEG2GYRBMPN7XGPY4OJNUJIFW/action/replication_record"}},"created_at":"2026-07-05T06:09:48.510406+00:00","updated_at":"2026-07-05T06:09:48.510406+00:00"}