Reinforcement Learning with Large Action Spaces for Neural Machine Translation
This addresses a bottleneck in RL for NMT, offering an incremental improvement for machine translation systems.
The paper tackles the problem that reinforcement learning (RL) for neural machine translation (NMT) is ineffective due to large action spaces, and finds that reducing the action space dimension improves performance, achieving an average gain of 1.5 BLEU points.
Applying Reinforcement learning (RL) following maximum likelihood estimation (MLE) pre-training is a versatile method for enhancing neural machine translation (NMT) performance. However, recent work has argued that the gains produced by RL for NMT are mostly due to promoting tokens that have already received a fairly high probability in pre-training. We hypothesize that the large action space is a main obstacle to RL's effectiveness in MT, and conduct two sets of experiments that lend support to our hypothesis. First, we find that reducing the size of the vocabulary improves RL's effectiveness. Second, we find that effectively reducing the dimension of the action space without changing the vocabulary also yields notable improvement as evaluated by BLEU, semantic similarity, and human evaluation. Indeed, by initializing the network's final fully connected layer (that maps the network's internal dimension to the vocabulary dimension), with a layer that generalizes over similar actions, we obtain a substantial improvement in RL performance: 1.5 BLEU points on average.