{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "oh-singh-lee-value-prediction-network-2017",
    "title": "Value Prediction Networks: planning with abstract rewards and values",
    "summary": "An observation-based planner tries to predict what it will see after an action. A Value Prediction Network instead learns action-option-dependent changes in an abstract state and predicts rewards, discounts and future values. It can expand several possible options and back up their estimated returns without generating future images. Oh, Singh and Lee combine this lookahead with multistep learning and evaluate it in a collection task and selected Atari games. This adds controlled planning to the abstract-value approach, rather than re-awarding the earlier uncontrolled Predictron.",
    "boundary": "The search uses a restricted branching heuristic and hand-designed options. An abstract state is not proved to be an exact representation of every relevant future. Deeper planning beyond the training range is not guaranteed to help. The Atari comparison improves seven of nine games and uses selected top agents, with a frame-skip setting that affects comparisons. These are finite results, not an optimal-policy theorem, universal noise robustness or the later MuZero system.",
    "area": "Model-based reinforcement learning",
    "topic": "Artificial intelligence",
    "year": 2017,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "oh-singh-lee-value-prediction-network-2017",
    "target": {
      "record_id": "urn:uuid:35d65c1f-22dc-47f9-a9b0-fc3a2c7717f9",
      "version": 1,
      "content_hash": "sha256:ec352f41ae9f393b61295ab833f609d4d9bac2f9357b9678ddce4a4e9bdb6aab"
    },
    "humans": [
      "Junhyuk Oh",
      "Satinder Singh",
      "Honglak Lee"
    ],
    "authors_complete": true,
    "human_role_note": "Source-listed human researchers and engineers for the exact method or resource. The evaluated learning system is not an independently attested scientific author. Earlier methods, funding and data-collection roles remain separate. Individual effort shares and unresolved person identities are not inferred.",
    "agents": [],
    "related": [
      "deep-q-network"
    ],
    "subresults": [],
    "scientific_assessment": {
      "source_review": "SOURCE_CURATED",
      "scientific_review": "NOT_REVIEWED_BY_WRS",
      "formal_replay": "NOT_RUN",
      "provider_attestation": "NONE"
    },
    "sources": [
      {
        "id": "cur20261017-arxiv-1707-03497v1",
        "title": "Value Prediction Network",
        "url": "https://arxiv.org/abs/1707.03497v1",
        "date": "2017-07-11"
      }
    ],
    "points": "65000",
    "published_grade": {
      "id": "oh-singh-lee-value-prediction-network-2017",
      "entry_id": "oh-singh-lee-value-prediction-network-2017",
      "band": "A",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:35d65c1f-22dc-47f9-a9b0-fc3a2c7717f9",
        "version": 1,
        "content_hash": "sha256:ec352f41ae9f393b61295ab833f609d4d9bac2f9357b9678ddce4a4e9bdb6aab"
      },
      "what_changed": "An observation-based planner tries to predict what it will see after an action. A Value Prediction Network instead learns action-option-dependent changes in an abstract state and predicts rewards, discounts and future values. It can expand several possible options and back up their estimated returns without generating future images. Oh, Singh and Lee combine this lookahead with multistep learning and evaluate it in a collection task and selected Atari games. This adds controlled planning to the abstract-value approach, rather than re-awarding the earlier uncontrolled Predictron.",
      "scope": "An observation-based planner tries to predict what it will see after an action. A Value Prediction Network instead learns action-option-dependent changes in an abstract state and predicts rewards, discounts and future values. It can expand several possible options and back up their estimated returns without generating future images. Oh, Singh and Lee combine this lookahead with multistep learning and evaluate it in a collection task and selected Atari games. This adds controlled planning to the abstract-value approach, rather than re-awarding the earlier uncontrolled Predictron. The search uses a restricted branching heuristic and hand-designed options. An abstract state is not proved to be an exact representation of every relevant future. Deeper planning beyond the training range is not guaranteed to help. The Atari comparison improves seven of nine games and uses selected top agents, with a frame-skip setting that affects comparisons. These are finite results, not an optimal-policy theorem, universal noise robustness or the later MuZero system.",
      "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:35d65c1f-22dc-47f9-a9b0-fc3a2c7717f9",
        "version": 1,
        "content_hash": "sha256:ec352f41ae9f393b61295ab833f609d4d9bac2f9357b9678ddce4a4e9bdb6aab"
      },
      "edition": "ratings-e039106f30dec177"
    }
  },
  "version": {
    "kind": "PUBLIC_RECORD_VIEW",
    "previous": null,
    "published_at": "2026-10-23T00:00:00Z",
    "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": "Junhyuk Oh",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Value Prediction Network",
            "url": "https://arxiv.org/abs/1707.03497v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Satinder Singh",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Value Prediction Network",
            "url": "https://arxiv.org/abs/1707.03497v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Honglak Lee",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Value Prediction Network",
            "url": "https://arxiv.org/abs/1707.03497v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "Value Prediction Network",
            "url": "https://arxiv.org/abs/1707.03497v1"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "The search uses a restricted branching heuristic and hand-designed options. An abstract state is not proved to be an exact representation of every relevant future. Deeper planning beyond the training range is not guaranteed to help. The Atari comparison improves seven of nine games and uses selected top agents, with a frame-skip setting that affects comparisons. These are finite results, not an optimal-policy theorem, universal noise robustness or the later MuZero system.",
        "domain": "Model-based reinforcement learning"
      },
      "sources": [
        {
          "citation": "Value Prediction Network",
          "url": "https://arxiv.org/abs/1707.03497v1"
        }
      ],
      "statement": "# Value Prediction Networks: planning with abstract rewards and values\n\nAn observation-based planner tries to predict what it will see after an action. A Value Prediction Network instead learns action-option-dependent changes in an abstract state and predicts rewards, discounts and future values. It can expand several possible options and back up their estimated returns without generating future images. Oh, Singh and Lee combine this lookahead with multistep learning and evaluate it in a collection task and selected Atari games. This adds controlled planning to the abstract-value approach, rather than re-awarding the earlier uncontrolled Predictron.\n\n## Scientific scope\n\nThe search uses a restricted branching heuristic and hand-designed options. An abstract state is not proved to be an exact representation of every relevant future. Deeper planning beyond the training range is not guaranteed to help. The Atari comparison improves seven of nine games and uses selected top agents, with a frame-skip setting that affects comparisons. These are finite results, not an optimal-policy theorem, universal noise robustness or the later MuZero system.\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\nJunhyuk Oh; Satinder Singh; Honglak Lee\n\nSource-listed human researchers and engineers for the exact method or resource. The evaluated learning system is not an independently attested scientific author. Earlier methods, funding and data-collection roles remain separate. Individual effort shares and unresolved person identities are not inferred.\n\n\n\n## Original sources\n\n- [Value Prediction Network](https://arxiv.org/abs/1707.03497v1)\n",
      "summary": "An observation-based planner tries to predict what it will see after an action. A Value Prediction Network instead learns action-option-dependent changes in an abstract state and predicts rewards, discounts and future values. It can expand several possible options and back up their estimated returns without generating future images. Oh, Singh and Lee combine this lookahead with multistep learning and evaluate it in a collection task and selected Atari games. This adds controlled planning to the abstract-value approach, rather than re-awarding the earlier uncontrolled Predictron.",
      "title": "Value Prediction Networks: planning with abstract rewards and values"
    },
    "record_id": "urn:uuid:35d65c1f-22dc-47f9-a9b0-fc3a2c7717f9",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:35d65c1f-22dc-47f9-a9b0-fc3a2c7717f9",
      "version": 1,
      "content_hash": "sha256:ec352f41ae9f393b61295ab833f609d4d9bac2f9357b9678ddce4a4e9bdb6aab"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:e220e4ef22b2ee8e7e8d4ed9dfa9a8d7b45ac07cfdb5467bc340c31d0edbe51e"
  },
  "target": {
    "record_id": "urn:uuid:35d65c1f-22dc-47f9-a9b0-fc3a2c7717f9",
    "version": 1,
    "content_hash": "sha256:ec352f41ae9f393b61295ab833f609d4d9bac2f9357b9678ddce4a4e9bdb6aab"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "oh-singh-lee-value-prediction-network-2017",
    "entry_id": "oh-singh-lee-value-prediction-network-2017",
    "revision": 1,
    "title": "Value Prediction Networks: planning with abstract rewards and values",
    "context": "Model-based reinforcement learning",
    "track": "Reusable scientific method, experimental result or historical framework",
    "band": "A",
    "modifier": null,
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-23",
    "evidence_cutoff": "2026-10-23",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "An observation-based planner tries to predict what it will see after an action. A Value Prediction Network instead learns action-option-dependent changes in an abstract state and predicts rewards, discounts and future values. It can expand several possible options and back up their estimated returns without generating future images. Oh, Singh and Lee combine this lookahead with multistep learning and evaluate it in a collection task and selected Atari games. This adds controlled planning to the abstract-value approach, rather than re-awarding the earlier uncontrolled Predictron.",
    "scope": "An observation-based planner tries to predict what it will see after an action. A Value Prediction Network instead learns action-option-dependent changes in an abstract state and predicts rewards, discounts and future values. It can expand several possible options and back up their estimated returns without generating future images. Oh, Singh and Lee combine this lookahead with multistep learning and evaluate it in a collection task and selected Atari games. This adds controlled planning to the abstract-value approach, rather than re-awarding the earlier uncontrolled Predictron. The search uses a restricted branching heuristic and hand-designed options. An abstract state is not proved to be an exact representation of every relevant future. Deeper planning beyond the training range is not guaranteed to help. The Atari comparison improves seven of nine games and uses selected top agents, with a frame-skip setting that affects comparisons. These are finite results, not an optimal-policy theorem, universal noise robustness or the later MuZero system.",
    "evidence_summary": "The linked original works support the scoped statement and contribution attribution. WRS has not independently replayed experiments or reconstructed a full mathematical proof. The search uses a restricted branching heuristic and hand-designed options. An abstract state is not proved to be an exact representation of every relevant future. Deeper planning beyond the training range is not guaranteed to help. The Atari comparison improves seven of nine games and uses selected top agents, with a frame-skip setting that affects comparisons. These are finite results, not an optimal-policy theorem, universal noise robustness or the later MuZero system.",
    "references": [
      {
        "title": "Value Prediction Network",
        "url": "https://arxiv.org/abs/1707.03497v1",
        "locator": "Original scoped contribution, statement/method and evidence limits; source-specific attribution retained."
      }
    ],
    "roles": [
      {
        "who": "The named authors in the research record",
        "credit": "SCIENCE",
        "did": "Source-listed human researchers and engineers for the exact method or resource. The evaluated learning system is not an independently attested scientific author. Earlier methods, funding and data-collection roles remain separate. Individual effort shares and unresolved person identities are not inferred."
      }
    ],
    "target": {
      "record_id": "urn:uuid:35d65c1f-22dc-47f9-a9b0-fc3a2c7717f9",
      "version": 1,
      "content_hash": "sha256:ec352f41ae9f393b61295ab833f609d4d9bac2f9357b9678ddce4a4e9bdb6aab"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "65000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}