CUED Publications database

Sequence Tutor: Conservative Fine-Tuning of Sequence Generation Models with KL-control

Jaques, N and Gu, S and Bahdanau, D and Hernández-Lobato, JM and Turner, RE and Eck, D Sequence Tutor: Conservative Fine-Tuning of Sequence Generation Models with KL-control. (Unpublished)

Full text not available from this repository.

Abstract

This paper proposes a general method for improving the structure and quality of sequences generated by a recurrent neural network (RNN), while maintaining information originally learned from data, as well as sample diversity. An RNN is first pre-trained on data using maximum likelihood estimation (MLE), and the probability distribution over the next token in the sequence learned by this model is treated as a prior policy. Another RNN is then trained using reinforcement learning (RL) to generate higher-quality outputs that account for domain-specific incentives while retaining proximity to the prior policy of the MLE RNN. To formalize this objective, we derive novel off-policy RL methods for RNNs from KL-control. The effectiveness of the approach is demonstrated on two applications; 1) generating novel musical melodies, and 2) computational molecular generation. For both problems, we show that the proposed method improves the desired properties and structure of the generated sequences, while maintaining information learned from data.

Item Type: Article
Uncontrolled Keywords: cs.LG cs.LG cs.AI
Subjects: UNSPECIFIED
Divisions: Div F > Computational and Biological Learning
Depositing User: Cron Job
Date Deposited: 17 Jul 2017 20:01
Last Modified: 27 Jul 2017 05:27
DOI: