{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "ai-oct-smoothgrad",
    "title": "SmoothGrad: noise-averaged gradient sensitivity maps",
    "summary": "Averages sensitivity maps over Gaussian perturbations of an input to reduce visually noisy saliency, and studies visualization choices and training-time noise.",
    "boundary": "Visual coherence is evaluated on selected image models and does not certify causal faithfulness, completeness or robustness. Monte Carlo averaging approximates a smoothed map, not the exact unsmoothed gradient.",
    "area": "AI interpretability, model editing and data attribution",
    "topic": "Artificial intelligence",
    "year": 2017,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "ai-oct-smoothgrad",
    "target": {
      "record_id": "urn:uuid:5c8d30a1-6a03-4d56-acbb-b7a2e45ce960",
      "version": 1,
      "content_hash": "sha256:2ffec0b098b76cfbbe057839e4a2ae276b80cfc06a967f8c8132be803585bd5b"
    },
    "humans": [
      "Daniel Smilkov",
      "Nikhil Thorat",
      "Been Kim",
      "Fernanda Viégas",
      "Martin Wattenberg"
    ],
    "authors_complete": true,
    "human_role_note": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.",
    "agents": [],
    "related": [],
    "subresults": [],
    "scientific_assessment": {
      "source_review": "SOURCE_CURATED",
      "scientific_review": "NOT_REVIEWED_BY_WRS",
      "formal_replay": "NOT_RUN",
      "provider_attestation": "NONE"
    },
    "sources": [
      {
        "id": "src-ai-oct-smoothgrad-1706-03825v1",
        "title": "SmoothGrad: removing noise by adding noise",
        "url": "https://arxiv.org/pdf/1706.03825v1"
      }
    ],
    "points": "130000",
    "published_grade": {
      "id": "ai-oct-smoothgrad",
      "entry_id": "ai-oct-smoothgrad",
      "band": "A",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:5c8d30a1-6a03-4d56-acbb-b7a2e45ce960",
        "version": 1,
        "content_hash": "sha256:2ffec0b098b76cfbbe057839e4a2ae276b80cfc06a967f8c8132be803585bd5b"
      },
      "what_changed": "Averages sensitivity maps over Gaussian perturbations of an input to reduce visually noisy saliency, and studies visualization choices and training-time noise.",
      "scope": "Averages sensitivity maps over Gaussian perturbations of an input to reduce visually noisy saliency, and studies visualization choices and training-time noise. Visual coherence is evaluated on selected image models and does not certify causal faithfulness, completeness or robustness. Monte Carlo averaging approximates a smoothed map, not the exact unsmoothed gradient.",
      "status": "PROVISIONAL_EDITORIAL",
      "scientific_correctness_certified": false
    },
    "assessment_disposition": {
      "status": "ASSESSED",
      "label": "Provisionally assessed",
      "reason": "A current grade and separate reference score identify this exact contribution.",
      "target": {
        "record_id": "urn:uuid:5c8d30a1-6a03-4d56-acbb-b7a2e45ce960",
        "version": 1,
        "content_hash": "sha256:2ffec0b098b76cfbbe057839e4a2ae276b80cfc06a967f8c8132be803585bd5b"
      },
      "edition": "ratings-e039106f30dec177"
    }
  },
  "version": {
    "kind": "PUBLIC_RECORD_VIEW",
    "previous": null,
    "published_at": "2026-10-16T00:00:00Z",
    "publisher": {
      "principal_id": "urn:uuid:80116931-785b-40a2-a24b-9d59db0c7e33",
      "role": "WRS_EDITORIAL_PUBLICATION"
    },
    "record": {
      "contributors": [
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Daniel Smilkov",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "SmoothGrad: removing noise by adding noise",
            "url": "https://arxiv.org/pdf/1706.03825v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Nikhil Thorat",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "SmoothGrad: removing noise by adding noise",
            "url": "https://arxiv.org/pdf/1706.03825v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Been Kim",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "SmoothGrad: removing noise by adding noise",
            "url": "https://arxiv.org/pdf/1706.03825v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Fernanda Viégas",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "SmoothGrad: removing noise by adding noise",
            "url": "https://arxiv.org/pdf/1706.03825v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Martin Wattenberg",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "SmoothGrad: removing noise by adding noise",
            "url": "https://arxiv.org/pdf/1706.03825v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "SmoothGrad: removing noise by adding noise",
            "url": "https://arxiv.org/pdf/1706.03825v1"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "Visual coherence is evaluated on selected image models and does not certify causal faithfulness, completeness or robustness. Monte Carlo averaging approximates a smoothed map, not the exact unsmoothed gradient.",
        "domain": "AI interpretability, model editing and data attribution"
      },
      "sources": [
        {
          "citation": "SmoothGrad: removing noise by adding noise",
          "url": "https://arxiv.org/pdf/1706.03825v1"
        }
      ],
      "statement": "# SmoothGrad: noise-averaged gradient sensitivity maps\n\nAverages sensitivity maps over Gaussian perturbations of an input to reduce visually noisy saliency, and studies visualization choices and training-time noise.\n\n## Scientific scope\n\nVisual coherence is evaluated on selected image models and does not certify causal faithfulness, completeness or robustness. Monte Carlo averaging approximates a smoothed map, not the exact unsmoothed gradient.\n\n## Reading status\n\nThis is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.\n\n## People and systems\n\nDaniel Smilkov; Nikhil Thorat; Been Kim; Fernanda Viégas; Martin Wattenberg\n\nSource-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.\n\n\n\n## Original sources\n\n- [SmoothGrad: removing noise by adding noise](https://arxiv.org/pdf/1706.03825v1)\n",
      "summary": "Averages sensitivity maps over Gaussian perturbations of an input to reduce visually noisy saliency, and studies visualization choices and training-time noise.",
      "title": "SmoothGrad: noise-averaged gradient sensitivity maps"
    },
    "record_id": "urn:uuid:5c8d30a1-6a03-4d56-acbb-b7a2e45ce960",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:5c8d30a1-6a03-4d56-acbb-b7a2e45ce960",
      "version": 1,
      "content_hash": "sha256:2ffec0b098b76cfbbe057839e4a2ae276b80cfc06a967f8c8132be803585bd5b"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:2775f1094182e36580cc5f1ca59fc18346aa6739a797185245420d4bf8578c25"
  },
  "target": {
    "record_id": "urn:uuid:5c8d30a1-6a03-4d56-acbb-b7a2e45ce960",
    "version": 1,
    "content_hash": "sha256:2ffec0b098b76cfbbe057839e4a2ae276b80cfc06a967f8c8132be803585bd5b"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "ai-oct-smoothgrad",
    "entry_id": "ai-oct-smoothgrad",
    "revision": 1,
    "title": "SmoothGrad: noise-averaged gradient sensitivity maps",
    "context": "AI interpretability, model editing and data attribution",
    "track": "Human-designed research method or resource",
    "band": "A",
    "modifier": null,
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-16",
    "evidence_cutoff": "2026-10-16",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "Averages sensitivity maps over Gaussian perturbations of an input to reduce visually noisy saliency, and studies visualization choices and training-time noise.",
    "scope": "Averages sensitivity maps over Gaussian perturbations of an input to reduce visually noisy saliency, and studies visualization choices and training-time noise. Visual coherence is evaluated on selected image models and does not certify causal faithfulness, completeness or robustness. Monte Carlo averaging approximates a smoothed map, not the exact unsmoothed gradient.",
    "evidence_summary": "The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Visual coherence is evaluated on selected image models and does not certify causal faithfulness, completeness or robustness. Monte Carlo averaging approximates a smoothed map, not the exact unsmoothed gradient.",
    "references": [
      {
        "title": "SmoothGrad: removing noise by adding noise",
        "url": "https://arxiv.org/pdf/1706.03825v1",
        "locator": "Original contribution and stated assumptions, evaluation scope and limitations; source-linked assessment, not independent execution or proof certification."
      }
    ],
    "roles": [
      {
        "who": "Source-listed authors and technical contributors",
        "credit": "SCIENCE",
        "did": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities."
      }
    ],
    "target": {
      "record_id": "urn:uuid:5c8d30a1-6a03-4d56-acbb-b7a2e45ce960",
      "version": 1,
      "content_hash": "sha256:2ffec0b098b76cfbbe057839e4a2ae276b80cfc06a967f8c8132be803585bd5b"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "130000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}