Faster Machine Translation Ensembling with Reinforcement Learning and Competitive Correction

Kritarth Prasad, Mohammadi Zaki, Pratik Rakesh Singh, Pankaj Wasnik

Abstract:

Ensembling neural machine translation (NMT)
models to produce higher-quality translations
than the Lindividual models has been exten-
sively studied. Recent methods typically em-
ploy a candidate selection block (CSB) and an
encoder-decoder fusion block (FB), requiring
inference across all candidate models, leading
to significant computational overhead, gener-
ally Ω(L). This paper introduces SmartGen,
a reinforcement learning (RL)-based strategy
that improves the CSB by selecting a small,
fixed number of candidates and identifying op-
timal groups to pass to the fusion block for
each input sentence. Furthermore, previously,
the CSB and FB were trained independently,
leading to suboptimal NMT performance. Our
DQN-based SmartGen addresses this by using
feedback from the FB block as a reward during
training. We also resolve a key issue in ear-
lier methods, where candidates were passed to
the FB without modification, by introducing a
Competitive Correction Block (CCB). Finally,
we validate our approach with extensive exper-
iments on English-Hindi translation tasks in
both directions as well as English to Chinese
and English to German.