Optimizing human-interpretable dialog management policy using Genetic Algorithm

TL;DR

This paper introduces a genetic algorithm-based framework for optimizing human-interpretable dialog management policies in spoken dialog systems, enhancing practical usability and robustness to noise.

Contribution

It proposes a novel method to optimize dialog policies in domain language, making them understandable and easier to verify or modify by system designers.

Findings

Effective optimization of dialog policies demonstrated

Applicable to both simulated and real human-machine dialogs

Improves practical deployment of dialog systems

Abstract

Automatic optimization of spoken dialog management policies that are robust to environmental noise has long been the goal for both academia and industry. Approaches based on reinforcement learning have been proved to be effective. However, the numerical representation of dialog policy is human-incomprehensible and difficult for dialog system designers to verify or modify, which limits its practical application. In this paper we propose a novel framework for optimizing dialog policies specified in domain language using genetic algorithm. The human-interpretable representation of policy makes the method suitable for practical employment. We present learning algorithms using user simulation and real human-machine dialogs respectively.Empirical experimental results are given to show the effectiveness of the proposed approach.