We present MILABOT: a deep reinforcement learning chatbot developed by the Montreal Institute for Learning Algorithms (MILA) for the Amazon Alexa Prize competition. MILABOT is capable of conversing with humans on popular small talk topics through both speech and text. The system consists of an ensemble of natural language generation and retrieval models, including neural network and template-based models. By applying reinforcement learning to crowdsourced data and real-world user interactions, the system has been trained to select an appropriate response from the models in its ensemble. The system has been evaluated through A/B testing with real-world users, where it performed significantly better than other systems. The results highlight the potential of coupling ensemble systems with deep reinforcement learning as a fruitful path for developing real-world, open-domain conversational agents.
@article{arxiv.1801.06700,
title = {A Deep Reinforcement Learning Chatbot (Short Version)},
author = {Iulian V. Serban and Chinnadhurai Sankar and Mathieu Germain and Saizheng Zhang and Zhouhan Lin and Sandeep Subramanian and Taesup Kim and Michael Pieper and Sarath Chandar and Nan Rosemary Ke and Sai Rajeswar and Alexandre de Brebisson and Jose M. R. Sotelo and Dendi Suhubdy and Vincent Michalski and Alexandre Nguyen and Joelle Pineau and Yoshua Bengio},
journal= {arXiv preprint arXiv:1801.06700},
year = {2018}
}
Comments
9 pages, 1 figure, 2 tables; presented at NIPS 2017, Conversational AI: "Today's Practice and Tomorrow's Potential" Workshop