{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.02396", "kind": "paper", "url": "https://arxiv.org/abs/2608.02396"}, {"id": "gh:nishan-charlie/vit_xai_bench", "kind": "repo", "url": "https://github.com/Nishan-Charlie/VIT_XAI_Bench"}], "authors": ["Sathiyamohan Nishankar", "Nethmi Pathirana", "Pubudu Sanjeewani", "Asanka Perera", "Selvarajah Thuseethan"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.02396", "source_url": "https://arxiv.org/abs/2608.02396"}, "categories": ["General AI", "Robustness", "Safety & Trustworthiness", "cs.CV"], "description": {"en": "A controlled evaluation of 13 attribution methods across CNN and Vision Transformer architectures using five quality and cost dimensions."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.02396", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.02396", "title": "Does Explainability Transfer? A Controlled Benchmark of Attribution Methods on Vision Transformers and CNNs"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.02396", "modality": null, "name": "Does Explainability Transfer? A Controlled Benchmark of Attribution Methods on Vision Transformers and CNNs", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-09-25T04:40:22Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_does-explainability-transfer-a-controlled-_ef895886", "origin_source": "arxiv", "recognition_confidence": "0.7", "record_sha256": "3c32d06f2d4be0f9b1a0b1699d8eacbaddc57c0763b00cdc473f97f95c30a14c", "relation": "introduces", "review_model": "codex-cli-default", "review_state": "ai-reviewed", "reviewed_at": "2026-09-24T17:36:59.562667Z", "source_url": "https://arxiv.org/abs/2608.02396"}, "publisher": null, "released": "2026-08-03", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.02396", "source_url": "https://arxiv.org/abs/2608.02396"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-02396", "source": "claire_radar", "source_benchmark_id": "2608.02396", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Safety & Trustworthiness", "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "The cross-architecture XAI comparison may attract specialist interest, though its likely reach is concentrated in explainability and computer-vision research.", "score": 64}, "capabilities": ["Robustness"], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-24T17:36:59.562667Z", "inputHash": "b062f76866ad00e9fe4cd1222b22eef12df2925a1fbd1e06ebd0d8d31f53aa4f", "model": "codex-cli-default"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "Does Explainability Transfer? A Controlled Benchmark of Attribution Methods on Vision Transformers and CNNs", "canonicalNameSource": "paper_title", "decisionReason": "The paper names a controlled benchmark, specifies its evaluation dimensions, and provides an open-source evaluation framework through the linked code repository.", "evaluationEvidence": {"reuse": "The open-source code for the evaluation framework and benchmark results is available", "scoring": "five dimensions: faithfulness, localization, robustness, complexity, and computational cost", "task": "evaluates attribution quality"}, "model": "codex-cli-default", "reviewedAt": "2026-09-24T17:36:59.562667Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A controlled evaluation of 13 attribution methods across CNN and Vision Transformer architectures using five quality and cost dimensions.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["benchmark term in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "This paper presents a controlled benchmark that evaluates attribution quality across five dimensions: faithfulness, localization, robustness, complexity, and computational cost."}, "familyId": "bmf_0a733b874ffa", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_does-explainability-transfer-a-controlled-_ef895886", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/Nishan-Charlie/VIT_XAI_Bench", "data": null, "paper": "https://arxiv.org/abs/2608.02396", "pdf": "https://arxiv.org/pdf/2608.02396", "project": null, "report": "https://arxiv.org/abs/2608.02396"}, "metrics": [], "motivation": "Most evidence on the effectiveness of explainable artificial intelligence (XAI) attribution methods has been established on convolutional neural networks (CNNs), with limited investigation into whether these conclusions generalize to the diverse Vision Transformer (ViT) architectures that now dominate computer vision.", "name": "Does Explainability Transfer? A Controlled Benchmark of Attribution Methods on Vision Transformers and CNNs", "oneLine": "A controlled evaluation of 13 attribution methods across CNN and Vision Transformer architectures using five quality and cost dimensions.", "paperTitle": "Does Explainability Transfer? A Controlled Benchmark of Attribution Methods on Vision Transformers and CNNs", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.02396", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "readiness": "Runnable", "recognitionConfidence": 0.7, "relation": "introduces", "releasedAt": "2026-08-03", "reviewLifecycle": {"state": "pending"}, "source": {"authors": ["Sathiyamohan Nishankar", "Nethmi Pathirana", "Pubudu Sanjeewani", "Asanka Perera", "Selvarajah Thuseethan"], "categories": ["cs.CV"], "id": "2608.02396", "title": "Does Explainability Transfer? A Controlled Benchmark of Attribution Methods on Vision Transformers and CNNs", "type": "arxiv", "url": "https://arxiv.org/abs/2608.02396"}, "sourceUpdatedAt": "2026-08-03T15:39:36Z", "topics": ["cs.CV"], "venueAttempts": [], "whyItMatters": "It tests whether attribution-method rankings transfer across architecture families and exposes limitations of single-metric evaluation."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
