[1Z1] [General Session] 2. Machine Learning

[1Z1-03] Expert-based reward function training: the novel method to train sequence generators

〇Joji Toyama¹, Yusuke Iwasawa¹, Yutaka Matsuo¹ (1. The University of Tokyo)

Keywords:Sequence modelling

The training methods of sequence generator with a combination of GAN and policy gradient has shown good performance. In this paper, we propose expert-based reward function training: the novel method to train sequence generator. Different from previous studies of sequence generation, expert-based reward function training does not utilize GAN's framework. Still, our model outperforms SeqGAN and a strong baseline, RankGAN.