This paper studies fitted value iteration for continuous state numerical dynamic programming using nonexpansive function approximators. A number of approximation schemes are discussed. The main contribution is to provide error bounds for approximate optimal policies generated by the value iteration algorithm. Journal of Economic Literature Classifications: C61, C63

In this paper we study decision making models based on aggregation operators and, more specially, on Choquet integrals. The motivation of our work is to study the modeling capabilities of these operators and to build models that can approximate arbitrary functions. We describe and study two models that are universal approximators.

In this paper a linearly parameterized approximators based algorithm for identifying uncertain systems is proposed. The proposed algorithm ensures stability without previous knowledge of bounds for the “optimal” parameter, approximation error, and disturbances. This algorithm guarantees that the state error converges asymptotically to zero, even in the presence of uncertainties, as long as some...

Estimation of returns over time, the focus of temporal difference (TD) algorithms, imposes particular constraints on good function approximators or representations. Appropriate generalisation between states is determined by how similar their successors are, and representations should follow suit. This paper shows howTDmachinery can be used to learn such representations, and illustrates, using a...

A new paradigm for the implementation of the Hough Transform (HT) is presented in this paper. The paradigm makes use of the neural networks' properties as function approximators in order to avoid some problems of the standard HT implementation. Some encouraging results are presented.

Estimation of returns over time, the focus of temporal difference (TD) algorithms, imposes particular constraints on good function approximators or representations. Appropriate generalization between states is determined by how similar their successors are, and representations should follow suit. This paper shows how TD machinery can be used to learn such representations, and illustrates, using...