Partially Observable Markov Decision Process

A Partially Observable Markov Decision Process (POMDP) is a generalization of a Markov Decision Process. A POMDP models an agent decision process in which it is assumed that the system dynamics are determined by an MDP, but the agent cannot directly observe the underlying state. Instead, it must maintain a probability distribution over the set of possible states, based on a set of observations and observation probabilities, and the underlying MDP.

The POMDP framework is general enough to model a variety of real-world sequential decision processes. Applications include robot navigation problems, machine maintenance, and planning under uncertainty in general. The framework originated in the Operations Research community, and was later taken over by the Artificial Intelligence and Automated Planning communities.

An exact solution to a POMDP yields the optimal action for each possible belief over the world states. The optimal action maximizes (or minimizes) the expected reward (or cost) of the agent over a possibly infinite horizon. The sequence of optimal actions is known as the optimal policy of the agent for interacting with its environment.

Read more about Partially Observable Markov Decision Process:  Belief Update, Belief MDP, Approximate POMDP Solutions, POMDP Uses

Famous quotes containing the words partially, observable, decision and/or process:

    Some men have a necessity to be mean, as if they were exercising a faculty which they had to partially neglect since early childhood.
    F. Scott Fitzgerald (1896–1940)

    Every living language, like the perspiring bodies of living creatures, is in perpetual motion and alteration; some words go off, and become obsolete; others are taken in, and by degrees grow into common use; or the same word is inverted to a new sense or notion, which in tract of time makes an observable change in the air and features of a language, as age makes in the lines and mien of a face.
    Richard Bentley (1662–1742)

    A good decision is based on knowledge and not on numbers.
    Plato (c. 427–347 B.C.)

    A designer who is not also a couturier, who hasn’t learned the most refined mysteries of physically creating his models, is like a sculptor who gives his drawings to another man, an artisan, to accomplish. For him the truncated process of creating will always be an interrupted act of love, and his style will bear the shame of it, the impoverishment.
    Yves Saint Laurent (b. 1936)