# Amortized Noisy Channel Neural Machine Translation

arXiv: https://arxiv.org/abs/2112.08670 · HTML: https://mlsanity.com/paper/2112.08670/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 29 of its 46 references that are in the map.
- First posted on arXiv 16 Dec 2021.

## Builds on (most cited first, 29 of 29)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [A Call for Clarity in Reporting BLEU Scores](https://mlsanity.com/paper/1804.08771/) (2018)
- [fairseq: A Fast, Extensible Toolkit for Sequence Modeling](https://mlsanity.com/paper/1904.01038/) (2019)
- [BLEURT: Learning Robust Metrics for Text Generation](https://mlsanity.com/paper/2004.04696/) (2020)
- [Sequence-Level Knowledge Distillation](https://mlsanity.com/paper/1606.07947/) (2016)
- [Sequence Level Training with Recurrent Neural Networks](https://mlsanity.com/paper/1511.06732/) (2015)
- [Non-Autoregressive Neural Machine Translation](https://mlsanity.com/paper/1711.02281/) (2017)
- [Parallel WaveNet: Fast High-Fidelity Speech Synthesis](https://mlsanity.com/paper/1711.10433/) (2017)
- [Deterministic Non-Autoregressive Neural Sequence Modeling by Iterative Refinement](https://mlsanity.com/paper/1802.06901/) (2018)
- [Facebook FAIR's WMT19 News Translation Task Submission](https://mlsanity.com/paper/1907.06616/) (2019)
- [An Actor-Critic Algorithm for Sequence Prediction](https://mlsanity.com/paper/1607.07086/) (2016)
- [Multilingual Neural Machine Translation with Knowledge Distillation](https://mlsanity.com/paper/1902.10461/) (2019)
- [On the Weaknesses of Reinforcement Learning for Neural Machine Translation](https://mlsanity.com/paper/1907.01752/) (2019)
- [Reward Augmented Maximum Likelihood for Neural Structured Prediction](https://mlsanity.com/paper/1609.00150/) (2016)
- [Text Generation by Learning from Demonstrations](https://mlsanity.com/paper/2009.07839/) (2020)
- [Facebook AI WMT21 News Translation Task Submission](https://mlsanity.com/paper/2108.03265/) (2021)
- [Consistency of a Recurrent Language Model With Respect to Incomplete Decoding](https://mlsanity.com/paper/2002.02492/) (2020)
- [Ensemble Distillation for Neural Machine Translation](https://mlsanity.com/paper/1702.01802/) (2017)
- [Simple and Effective Noisy Channel Modeling for Neural Machine Translation](https://mlsanity.com/paper/1908.05731/) (2019)
- [A Deep Reinforcement Learning Chatbot](https://mlsanity.com/paper/1709.02349/) (2017)
- [ENGINE: Energy-Based Inference Networks for Non-Autoregressive Machine Translation](https://mlsanity.com/paper/2005.00850/) (2020)
- [Trainable Greedy Decoding for Neural Machine Translation](https://mlsanity.com/paper/1702.02429/) (2017)
- [End-to-End Offline Goal-Oriented Dialog Policy Learning via Policy Gradient](https://mlsanity.com/paper/1712.02838/) (2017)
- [Gender Coreference and Bias Evaluation at WMT 2020](https://mlsanity.com/paper/2010.06018/) (2020)
- [The Neural Noisy Channel](https://mlsanity.com/paper/1611.02554/) (2016)
- [Facebook AI's WMT20 News Translation Task Submission](https://mlsanity.com/paper/2011.08298/) (2020)
- [A Stable and Effective Learning Strategy for Trainable Greedy Decoding](https://mlsanity.com/paper/1804.07915/) (2018)
- [Language Models not just for Pre-training: Fast Online Neural Noisy Channel Modeling](https://mlsanity.com/paper/2011.07164/) (2020)

## Built on by (most cited first, 4 of 4)

- [Token Dropping for Efficient BERT Pretraining](https://mlsanity.com/paper/2203.13240/) (2022)
- [Reward Gaming in Conditional Text Generation](https://mlsanity.com/paper/2211.08714/) (2022)
- [Leveraging Implicit Feedback from Deployment Data in Dialogue](https://mlsanity.com/paper/2307.14117/) (2023)
- [Joint Generator-Ranker Learning for Natural Language Generation](https://mlsanity.com/paper/2206.13974/) (2022)

## Abstract (arXiv)

Noisy channel models have been especially effective in neural machine translation (NMT). However, recent approaches like "beam search and rerank" (BSR) incur significant computation overhead during inference, making real-world application infeasible. We aim to study if it is possible to build an amortized noisy channel NMT model such that when we do greedy decoding during inference, the translation accuracy matches that of BSR in terms of reward (based on the source-to-target log probability and the target-to-source log probability) and quality (based on BLEU and BLEURT). We attempt three approaches to train the new model: knowledge distillation, one-step-deviation imitation learning, and Q learning. The first approach obtains the noisy channel signal from a pseudo-corpus, and the latter two approaches aim to optimize toward a noisy-channel MT reward directly. For all three approaches, the generated translations fail to achieve rewards comparable to BSR, but the translation quality approximated by BLEU and BLEURT is similar to the quality of BSR-produced translations. Additionally, all three approaches speed up inference by 1-2 orders of magnitude.