{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:XHTZS34PWGAPSHHDQ3GDCK6EFS","short_pith_number":"pith:XHTZS34P","schema_version":"1.0","canonical_sha256":"b9e7996f8fb180f91ce386cc312bc42c8a702624f236e4bf0dc4a1ccd8143689","source":{"kind":"arxiv","id":"2012.00073","version":2},"attestation_state":"computed","paper":{"title":"TimeSHAP: Explaining Recurrent Models through Sequence Perturbations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Andr\\'e F. Cruz, Jo\\~ao Bento, M\\'ario A.T. Figueiredo, Pedro Bizarro, Pedro Saleiro","submitted_at":"2020-11-30T19:48:57Z","abstract_excerpt":"Although recurrent neural networks (RNNs) are state-of-the-art in numerous sequential decision-making tasks, there has been little research on explaining their predictions. In this work, we present TimeSHAP, a model-agnostic recurrent explainer that builds upon KernelSHAP and extends it to the sequential domain. TimeSHAP computes feature-, timestep-, and cell-level attributions. As sequences may be arbitrarily long, we further propose a pruning method that is shown to dramatically decrease both its computational cost and the variance of its attributions. We use TimeSHAP to explain the predicti"},"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":"2012.00073","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-11-30T19:48:57Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1f6a8c46df6f0dbad8845ee1d8e665230ca2e32d4ca33243b94cabb3e947c767","abstract_canon_sha256":"3a2fd8f90a1bf8cdf1e65dff437b7a8a6175183e5663654f5e3b6f350399785b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:52:27.491304Z","signature_b64":"p5NKjQ4IdTfObxWkNUADDC+AdoEP6Whyb6I6DhmXlM+DrFP+4eWVE/YP7NPBaaWF27X49hYrDo+5mpMt0rxdDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b9e7996f8fb180f91ce386cc312bc42c8a702624f236e4bf0dc4a1ccd8143689","last_reissued_at":"2026-07-05T02:52:27.490878Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:52:27.490878Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TimeSHAP: Explaining Recurrent Models through Sequence Perturbations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Andr\\'e F. Cruz, Jo\\~ao Bento, M\\'ario A.T. Figueiredo, Pedro Bizarro, Pedro Saleiro","submitted_at":"2020-11-30T19:48:57Z","abstract_excerpt":"Although recurrent neural networks (RNNs) are state-of-the-art in numerous sequential decision-making tasks, there has been little research on explaining their predictions. In this work, we present TimeSHAP, a model-agnostic recurrent explainer that builds upon KernelSHAP and extends it to the sequential domain. TimeSHAP computes feature-, timestep-, and cell-level attributions. As sequences may be arbitrarily long, we further propose a pruning method that is shown to dramatically decrease both its computational cost and the variance of its attributions. We use TimeSHAP to explain the predicti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2012.00073","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/2012.00073/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":"2012.00073","created_at":"2026-07-05T02:52:27.490939+00:00"},{"alias_kind":"arxiv_version","alias_value":"2012.00073v2","created_at":"2026-07-05T02:52:27.490939+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2012.00073","created_at":"2026-07-05T02:52:27.490939+00:00"},{"alias_kind":"pith_short_12","alias_value":"XHTZS34PWGAP","created_at":"2026-07-05T02:52:27.490939+00:00"},{"alias_kind":"pith_short_16","alias_value":"XHTZS34PWGAPSHHD","created_at":"2026-07-05T02:52:27.490939+00:00"},{"alias_kind":"pith_short_8","alias_value":"XHTZS34P","created_at":"2026-07-05T02:52:27.490939+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2506.13539","citing_title":"Direct optimization of the probability of lesion origin in proton treatment planning for low-grade glioma patients","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00645","citing_title":"From Prediction to Practice: A Task-Aware Evaluation Framework for Blood Glucose Forecasting","ref_index":39,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XHTZS34PWGAPSHHDQ3GDCK6EFS","json":"https://pith.science/pith/XHTZS34PWGAPSHHDQ3GDCK6EFS.json","graph_json":"https://pith.science/api/pith-number/XHTZS34PWGAPSHHDQ3GDCK6EFS/graph.json","events_json":"https://pith.science/api/pith-number/XHTZS34PWGAPSHHDQ3GDCK6EFS/events.json","paper":"https://pith.science/paper/XHTZS34P"},"agent_actions":{"view_html":"https://pith.science/pith/XHTZS34PWGAPSHHDQ3GDCK6EFS","download_json":"https://pith.science/pith/XHTZS34PWGAPSHHDQ3GDCK6EFS.json","view_paper":"https://pith.science/paper/XHTZS34P","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2012.00073&json=true","fetch_graph":"https://pith.science/api/pith-number/XHTZS34PWGAPSHHDQ3GDCK6EFS/graph.json","fetch_events":"https://pith.science/api/pith-number/XHTZS34PWGAPSHHDQ3GDCK6EFS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XHTZS34PWGAPSHHDQ3GDCK6EFS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XHTZS34PWGAPSHHDQ3GDCK6EFS/action/storage_attestation","attest_author":"https://pith.science/pith/XHTZS34PWGAPSHHDQ3GDCK6EFS/action/author_attestation","sign_citation":"https://pith.science/pith/XHTZS34PWGAPSHHDQ3GDCK6EFS/action/citation_signature","submit_replication":"https://pith.science/pith/XHTZS34PWGAPSHHDQ3GDCK6EFS/action/replication_record"}},"created_at":"2026-07-05T02:52:27.490939+00:00","updated_at":"2026-07-05T02:52:27.490939+00:00"}