Improving Beam Search by Removing Monotonic Constraint for Neural Machine Translation
Raphael Shu, Hideki Nakayama · 2018
To achieve high translation performance, neural machine translation models usually rely on the beam search algorithm for decoding sentences.The beam search finds good candidate translations by considering multiple hypotheses of translations simultaneously.However, as the algorithm searches in a monotonic left-to-right order, a hypothesis can not be revisited once it is discarded.We found such monotonicity forces the algorithm to sacrifice some decoding paths to explore new paths.As a result, the overall quality of the hypotheses selected by the algorithm is lower than expected.To mitigate this problem, we relax the monotonic constraint of the beam search by maintaining all found hypotheses in a single priority queue and using a universal score function for hypothesis selection.The proposed algorithm allows discarded hypotheses to be recovered in a later step.Despite its simplicity, we show that the proposed decoding algorithm enhances the quality of selected hypotheses and improve the translations even for highperformance models in English-Japanese translation task.