Few-Shot Generalization Across Dialogue Tasks

TL;DR

This paper introduces REDP, a novel dialogue policy embedding system with memory and attention, demonstrating superior few-shot generalization across dialogue tasks and achieving perfect accuracy on the bAbI benchmark.

Contribution

The paper presents REDP, a new recurrent embedding dialogue policy with memory and attention, enabling better transfer and few-shot learning in dialogue management.

Findings

REDP outperforms baseline LSTM on dialogue tasks.

Both REDP and baseline achieve 100% accuracy on bAbI dialogue.

REDP effectively generalizes to new dialogue domains.

Abstract

Machine-learning based dialogue managers are able to learn complex behaviors in order to complete a task, but it is not straightforward to extend their capabilities to new domains. We investigate different policies' ability to handle uncooperative user behavior, and how well expertise in completing one task (such as restaurant reservations) can be reapplied when learning a new one (e.g. booking a hotel). We introduce the Recurrent Embedding Dialogue Policy (REDP), which embeds system actions and dialogue states in the same vector space. REDP contains a memory component and attention mechanism based on a modified Neural Turing Machine, and significantly outperforms a baseline LSTM classifier on this task. We also show that both our architecture and baseline solve the bAbI dialogue task, achieving 100% test accuracy.