A Unified View of TD Algorithms; Introducing Full-Gradient TD and Equi-Gradient Descent TD

TL;DR

This paper unifies various TD algorithms under a gradient minimization framework and introduces two new algorithms, Full-gradient TD and EGD TD, that improve sample efficiency while maintaining gradient descent properties.

Contribution

It provides a unified theoretical view of TD algorithms and introduces two novel algorithms that enhance sample efficiency and computational properties.

Findings

Full-gradient TD generalizes iLSTD principles.

EGD TD reduces gradients via successive equi-gradient descents.

The new algorithms outperform traditional TD in sample utilization.

Abstract

This paper addresses the issue of policy evaluation in Markov Decision Processes, using linear function approximation. It provides a unified view of algorithms such as TD(lambda), LSTD(lambda), iLSTD, residual-gradient TD. It is asserted that they all consist in minimizing a gradient function and differ by the form of this function and their means of minimizing it. Two new schemes are introduced in that framework: Full-gradient TD which uses a generalization of the principle introduced in iLSTD, and EGD TD, which reduces the gradient by successive equi-gradient descents. These three algorithms form a new intermediate family with the interesting property of making much better use of the samples than TD while keeping a gradient descent scheme, which is useful for complexity issues and optimistic policy iteration.