# Rethinking Supervised Learning and Reinforcement Learning in Task-Oriented Dialogue Systems

arXiv: https://arxiv.org/abs/2009.09781 · HTML: https://mlsanity.com/paper/2009.09781/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 Sep 2020.

## Built on by (most cited first, 10 of 10)

- [EmoWOZ: A Large-Scale Corpus and Labelling Scheme for Emotion Recognition in Task-Oriented Dialogue Systems](https://mlsanity.com/paper/2109.04919/) (2021)
- [Conversational Recommender System and Large Language Model Are Made for Each Other in E-commerce Pre-sales Dialogue](https://mlsanity.com/paper/2310.14626/) (2023)
- [Transforming Human-Centered AI Collaboration: Redefining Embodied Agents Capabilities through Interactive Grounded Language Instructions](https://mlsanity.com/paper/2305.10783/) (2023)
- [U-NEED: A Fine-grained Dataset for User Needs-Centric E-commerce Conversational Recommendation](https://mlsanity.com/paper/2305.04774/) (2023)
- ["Think Before You Speak": Improving Multi-Action Dialog Policy by Planning Single-Action Dialogs](https://mlsanity.com/paper/2204.11481/) (2022)
- [Rewarding What Matters: Step-by-Step Reinforcement Learning for Task-Oriented Dialogue](https://mlsanity.com/paper/2406.14457/) (2024)
- [It Couldn't Help But Overhear: On the Limits of Modelling Meta-Communicative Grounding Acts with Supervised Learning](https://mlsanity.com/paper/2405.01139/) (2024)
- [IDAT: A Multi-Modal Dataset and Toolkit for Building and Evaluating Interactive Task-Solving Agents](https://mlsanity.com/paper/2407.08898/) (2024)
- [JoTR: A Joint Transformer and Reinforcement Learning Framework for Dialog Policy Learning](https://mlsanity.com/paper/2309.00230/) (2023)
- [Dialog Action-Aware Transformer for Dialog Policy Learning](https://mlsanity.com/paper/2309.02240/) (2023)

## Abstract (arXiv)

Dialogue policy learning for task-oriented dialogue systems has enjoyed great progress recently mostly through employing reinforcement learning methods. However, these approaches have become very sophisticated. It is time to re-evaluate it. Are we really making progress developing dialogue agents only based on reinforcement learning? We demonstrate how (1)~traditional supervised learning together with (2)~a simulator-free adversarial learning method can be used to achieve performance comparable to state-of-the-art RL-based methods. First, we introduce a simple dialogue action decoder to predict the appropriate actions. Then, the traditional multi-label classification solution for dialogue policy learning is extended by adding dense layers to improve the dialogue agent performance. Finally, we employ the Gumbel-Softmax estimator to alternatively train the dialogue agent and the dialogue reward model without using reinforcement learning. Based on our extensive experimentation, we can conclude the proposed methods can achieve more stable and higher performance with fewer efforts, such as the domain knowledge required to design a user simulator and the intractable parameter tuning in reinforcement learning. Our main goal is not to beat reinforcement learning with supervised learning, but to demonstrate the value of rethinking the role of reinforcement learning and supervised learning in optimizing task-oriented dialogue systems.