{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:7CPDXGPCIUK5IMKDL5HMVGUKCG","short_pith_number":"pith:7CPDXGPC","canonical_record":{"source":{"id":"2605.15154","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2026-05-14T17:51:09Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"5caac11b245ebc0605b8d86c6bd5b15dcf167d0b22a694a22549ae6e85dbe962","abstract_canon_sha256":"496512f8e82e9fca1f8683b2f6d58d5e054c29d67599295fe887d4eaa2239d43"},"schema_version":"1.0"},"canonical_sha256":"f89e3b99e24515d431435f4eca9a8a11a0643ee84f3a02a32d26224dcb0761f6","source":{"kind":"arxiv","id":"2605.15154","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15154","created_at":"2026-05-17T21:18:33Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15154v1","created_at":"2026-05-17T21:18:33Z"},{"alias_kind":"pith_short_12","alias_value":"7CPDXGPCIUK5","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"7CPDXGPCIUK5IMKD","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"7CPDXGPC","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:7CPDXGPCIUK5IMKDL5HMVGUKCG","target":"record","payload":{"canonical_record":{"source":{"id":"2605.15154","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2026-05-14T17:51:09Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"5caac11b245ebc0605b8d86c6bd5b15dcf167d0b22a694a22549ae6e85dbe962","abstract_canon_sha256":"496512f8e82e9fca1f8683b2f6d58d5e054c29d67599295fe887d4eaa2239d43"},"schema_version":"1.0"},"canonical_sha256":"f89e3b99e24515d431435f4eca9a8a11a0643ee84f3a02a32d26224dcb0761f6","receipt":{"kind":"pith_receipt","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.2","canonical_sha256":"f89e3b99e24515d431435f4eca9a8a11a0643ee84f3a02a32d26224dcb0761f6","last_reissued_at":"2026-05-17T21:57:18.769604Z","signature_status":"unsigned_v0","first_computed_at":"2026-05-17T21:40:25.447168Z"},"source_kind":"arxiv","source_id":"2605.15154","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T21:18:33Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"H0oTwIxgZu3meDadt/970cAmjvIK/7JLBf/Usiv/LbeDXFDeeLyai6KR6FlVgrrO9x6blRGVJ4XAj22aVCmfAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-20T09:35:38.978121Z"},"content_sha256":"dada1d0477e4b996a18c5537cb38fb34462e8011c30df1c1a094419ef3bb1e32","schema_version":"1.0","event_id":"sha256:dada1d0477e4b996a18c5537cb38fb34462e8011c30df1c1a094419ef3bb1e32"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:7CPDXGPCIUK5IMKDL5HMVGUKCG","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"RoSHAP: A Distributional Framework and Robust Metric for Stable Feature Attribution","license":"http://creativecommons.org/licenses/by/4.0/","headline":"RoSHAP summarizes the distribution of SHAP values to rank features by their activity, strength, and stability simultaneously.","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Boyu Jiang, Dawei Zhou, Feng Guo, Lanxin Xiang, Liang Shi, Youhui Ye","submitted_at":"2026-05-14T17:51:09Z","abstract_excerpt":"Feature attribution analysis is critical for interpreting machine learning models and supporting reliable data-driven decisions. However, feature attribution measures often exhibit stochastic variation: different train--test splits, random seeds, or model-fitting procedures can produce substantially different attribution values and feature rankings. This paper proposes a framework for incorporating stochastic nature of feature attribution and a robust attribution metric, RoSHAP, for stable feature ranking based on the SHAP metric. The proposed framework models the distribution of feature attri"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We show that, under mild regularity conditions, the aggregated feature attribution score is asymptotically Gaussian, which greatly reduces the computational cost of distribution estimation. The RoSHAP summarizes the distribution of SHAP into a robust feature-ranking criterion that simultaneously rewards features that are active, strong, and stable.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The mild regularity conditions under which the aggregated feature attribution score is asymptotically Gaussian.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"RoSHAP is a robust feature-ranking metric that summarizes the distributional properties of SHAP values via bootstrap resampling and asymptotic normality to reward active, strong, and stable features.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"RoSHAP summarizes the distribution of SHAP values to rank features by their activity, strength, and stability simultaneously.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"82f02704366d9500799ffb5218338ca15d42ae95730931fb82d36bcb17b1963d"},"source":{"id":"2605.15154","kind":"arxiv","version":1},"verdict":{"id":"cc38427a-4ead-4d7b-baaa-5a009c21752d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T02:50:53.812798Z","strongest_claim":"We show that, under mild regularity conditions, the aggregated feature attribution score is asymptotically Gaussian, which greatly reduces the computational cost of distribution estimation. The RoSHAP summarizes the distribution of SHAP into a robust feature-ranking criterion that simultaneously rewards features that are active, strong, and stable.","one_line_summary":"RoSHAP is a robust feature-ranking metric that summarizes the distributional properties of SHAP values via bootstrap resampling and asymptotic normality to reward active, strong, and stable features.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The mild regularity conditions under which the aggregated feature attribution score is asymptotically Gaussian.","pith_extraction_headline":"RoSHAP summarizes the distribution of SHAP values to rank features by their activity, strength, and stability simultaneously."},"references":{"count":26,"sample":[{"doi":"","year":2017,"title":"A unified approach to interpreting model predictions.Advances in neural information processing systems, 30","work_id":"a3d4b3ad-741f-42c0-9996-a9625ea12468","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"why should i trust you?","work_id":"0c59d0fa-7afc-4999-bc21-08a6357898a2","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Learning important features through propa- gating activation differences","work_id":"2323578f-5e5c-4847-a522-392cc35384d8","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Explainable ai-based deep- shap for mapping the multivariate relationships between regional neuroimaging biomarkers and cognition","work_id":"a73a375e-e2ae-402a-b8d3-5cd3d3a22eee","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Practical guide to shap analysis: Explaining supervised machine learning model predictions in drug development.Clinical and translational science, 17(11):e70056, 2024","work_id":"9cf37f85-d9b9-4038-9edf-2b815596d751","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":26,"snapshot_sha256":"1d39f5268a04fa21c0bb6b095a414093844fac3d83276bd0148043a99233a018","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"2d7487d622b556bf55a0d9f023016cbeb76dc06374448948169799f76ca93522"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"cc38427a-4ead-4d7b-baaa-5a009c21752d"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T21:57:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lRzxCfrSBNUY4FzAMl50STnW5CFNGIrUmNXYCX2FCS2vgXH7S3ct8KOiFXZKhdVeQca0QwDWhZG3NL+PXpFaCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-20T09:35:38.978643Z"},"content_sha256":"818ce686e9e1056bbc6fe4788b002f7dd3c5996893f2950d6b94c30e190c50c1","schema_version":"1.0","event_id":"sha256:818ce686e9e1056bbc6fe4788b002f7dd3c5996893f2950d6b94c30e190c50c1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/7CPDXGPCIUK5IMKDL5HMVGUKCG/bundle.json","state_url":"https://pith.science/pith/7CPDXGPCIUK5IMKDL5HMVGUKCG/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/7CPDXGPCIUK5IMKDL5HMVGUKCG/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-20T09:35:38Z","links":{"resolver":"https://pith.science/pith/7CPDXGPCIUK5IMKDL5HMVGUKCG","bundle":"https://pith.science/pith/7CPDXGPCIUK5IMKDL5HMVGUKCG/bundle.json","state":"https://pith.science/pith/7CPDXGPCIUK5IMKDL5HMVGUKCG/state.json","well_known_bundle":"https://pith.science/.well-known/pith/7CPDXGPCIUK5IMKDL5HMVGUKCG/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:7CPDXGPCIUK5IMKDL5HMVGUKCG","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"496512f8e82e9fca1f8683b2f6d58d5e054c29d67599295fe887d4eaa2239d43","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2026-05-14T17:51:09Z","title_canon_sha256":"5caac11b245ebc0605b8d86c6bd5b15dcf167d0b22a694a22549ae6e85dbe962"},"schema_version":"1.0","source":{"id":"2605.15154","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15154","created_at":"2026-05-17T21:18:33Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15154v1","created_at":"2026-05-17T21:18:33Z"},{"alias_kind":"pith_short_12","alias_value":"7CPDXGPCIUK5","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"7CPDXGPCIUK5IMKD","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"7CPDXGPC","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:818ce686e9e1056bbc6fe4788b002f7dd3c5996893f2950d6b94c30e190c50c1","target":"graph","created_at":"2026-05-17T21:57:18Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We show that, under mild regularity conditions, the aggregated feature attribution score is asymptotically Gaussian, which greatly reduces the computational cost of distribution estimation. The RoSHAP summarizes the distribution of SHAP into a robust feature-ranking criterion that simultaneously rewards features that are active, strong, and stable."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The mild regularity conditions under which the aggregated feature attribution score is asymptotically Gaussian."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"RoSHAP is a robust feature-ranking metric that summarizes the distributional properties of SHAP values via bootstrap resampling and asymptotic normality to reward active, strong, and stable features."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"RoSHAP summarizes the distribution of SHAP values to rank features by their activity, strength, and stability simultaneously."}],"snapshot_sha256":"82f02704366d9500799ffb5218338ca15d42ae95730931fb82d36bcb17b1963d"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"2d7487d622b556bf55a0d9f023016cbeb76dc06374448948169799f76ca93522"},"paper":{"abstract_excerpt":"Feature attribution analysis is critical for interpreting machine learning models and supporting reliable data-driven decisions. However, feature attribution measures often exhibit stochastic variation: different train--test splits, random seeds, or model-fitting procedures can produce substantially different attribution values and feature rankings. This paper proposes a framework for incorporating stochastic nature of feature attribution and a robust attribution metric, RoSHAP, for stable feature ranking based on the SHAP metric. The proposed framework models the distribution of feature attri","authors_text":"Boyu Jiang, Dawei Zhou, Feng Guo, Lanxin Xiang, Liang Shi, Youhui Ye","cross_cats":["cs.LG"],"headline":"RoSHAP summarizes the distribution of SHAP values to rank features by their activity, strength, and stability simultaneously.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2026-05-14T17:51:09Z","title":"RoSHAP: A Distributional Framework and Robust Metric for Stable Feature Attribution"},"references":{"count":26,"internal_anchors":0,"resolved_work":26,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"A unified approach to interpreting model predictions.Advances in neural information processing systems, 30","work_id":"a3d4b3ad-741f-42c0-9996-a9625ea12468","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"why should i trust you?","work_id":"0c59d0fa-7afc-4999-bc21-08a6357898a2","year":2016},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Learning important features through propa- gating activation differences","work_id":"2323578f-5e5c-4847-a522-392cc35384d8","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Explainable ai-based deep- shap for mapping the multivariate relationships between regional neuroimaging biomarkers and cognition","work_id":"a73a375e-e2ae-402a-b8d3-5cd3d3a22eee","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Practical guide to shap analysis: Explaining supervised machine learning model predictions in drug development.Clinical and translational science, 17(11):e70056, 2024","work_id":"9cf37f85-d9b9-4038-9edf-2b815596d751","year":2024}],"snapshot_sha256":"1d39f5268a04fa21c0bb6b095a414093844fac3d83276bd0148043a99233a018"},"source":{"id":"2605.15154","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T02:50:53.812798Z","id":"cc38427a-4ead-4d7b-baaa-5a009c21752d","model_set":{"reader":"grok-4.3"},"one_line_summary":"RoSHAP is a robust feature-ranking metric that summarizes the distributional properties of SHAP values via bootstrap resampling and asymptotic normality to reward active, strong, and stable features.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"RoSHAP summarizes the distribution of SHAP values to rank features by their activity, strength, and stability simultaneously.","strongest_claim":"We show that, under mild regularity conditions, the aggregated feature attribution score is asymptotically Gaussian, which greatly reduces the computational cost of distribution estimation. The RoSHAP summarizes the distribution of SHAP into a robust feature-ranking criterion that simultaneously rewards features that are active, strong, and stable.","weakest_assumption":"The mild regularity conditions under which the aggregated feature attribution score is asymptotically Gaussian."}},"verdict_id":"cc38427a-4ead-4d7b-baaa-5a009c21752d"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:dada1d0477e4b996a18c5537cb38fb34462e8011c30df1c1a094419ef3bb1e32","target":"record","created_at":"2026-05-17T21:18:33Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"496512f8e82e9fca1f8683b2f6d58d5e054c29d67599295fe887d4eaa2239d43","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2026-05-14T17:51:09Z","title_canon_sha256":"5caac11b245ebc0605b8d86c6bd5b15dcf167d0b22a694a22549ae6e85dbe962"},"schema_version":"1.0","source":{"id":"2605.15154","kind":"arxiv","version":1}},"canonical_sha256":"f89e3b99e24515d431435f4eca9a8a11a0643ee84f3a02a32d26224dcb0761f6","receipt":{"builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f89e3b99e24515d431435f4eca9a8a11a0643ee84f3a02a32d26224dcb0761f6","first_computed_at":"2026-05-17T21:40:25.447168Z","kind":"pith_receipt","last_reissued_at":"2026-05-17T21:57:18.769604Z","receipt_version":"0.2","signature_status":"unsigned_v0"},"source_id":"2605.15154","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:dada1d0477e4b996a18c5537cb38fb34462e8011c30df1c1a094419ef3bb1e32","sha256:818ce686e9e1056bbc6fe4788b002f7dd3c5996893f2950d6b94c30e190c50c1"],"state_sha256":"5c24939d482d39550bcf1d2a649e7228d123c7c02feada7b64c609f57967111c"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wUGXkODpexmY7MZm7hXhkigladcHM/FE0Ap3qwsCUo3RIQnvyU+JYzLsDLINOuZu43AUjAMryMZKxUpULFa9BA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-20T09:35:38.981049Z","bundle_sha256":"4c4076a7102bf4539aa608e7e41ab129e496831292b64daafd8ea1cba00f24e2"}}