Home / Curated / Model Families / Partially observable Markov decision process

Partially observable Markov decision process

Model Families AIMech:PartiallyObservableMarkovDecisionProcess DRAFT

Aliases: POMDP

A Markov decision model in which observations provide incomplete information about the current state.

Identity Notes & Concept Scope Model does not require state uncertainty, actions, transition probabilities, observations and rewards.

Claims & Grounded Evidence

1 Claim

Mechanism #mechanism

The model specifies state transitions and an observation distribution conditional on the resulting state and action.

supports Source: Planning and acting in partially observable stochastic domains · Locator: Kaelbling, Littman and Cassandra (1998), section 3.1, printed p. 105 / PDF p. 7: POMDP tuple and observation model

This record supports method identity and the stated mechanism only; no universal performance or biological efficacy claim is curated.
Proposed placement is an inference from the existing AIO hierarchy; maintainers must resolve task, algorithm and model scope before acceptance.

Mapped AIO Classes

AIO Target Class	Mapping Relation	Mapping Rationale
`AIO:Model` Model	broad_match	Model does not require state uncertainty, actions, transition probabilities, observations and rewards. Mapping is local curation, not an upstream subclass assertion.

1 Source

Source ID	Publication / Document	Type / Access	Reference & Checksum
`method-source`	Planning and acting in partially observable stochastic domains Leslie Pack Kaelbling, Michael L. Littman, Anthony R. Cassandra	primary paper full_text

Curator: Codex

Method: ai_assisted

Model: GPT-6

Created: 2026-10-11T03:01:34Z

Updated: 2026-10-11T03:05:44Z

Source File: data/curated/PartiallyObservableMarkovDecisionProcess.yaml