Dual Learning for Machine Translation

Abstract

While neural machine translation (NMT) is making good progress in the past two years, tens of millions of bilingual sentence pairs are needed for its training. However, human labeling is very costly. To tackle this training data bottleneck, we develop a dual-learning mechanism, which can enable an NMT system to automatically learn from unlabeled data through a dual-learning game. This mechanism is inspired by the following observation: any machine translation task has a dual task, e.g., English-to-French translation (primal) versus French-to-English translation (dual); the primal and dual tasks can form a closed loop, and generate informative feedback signals to train the translation models, even if without the involvement of a human labeler. In the dual-learning mechanism, we use one agent to represent the model for the primal task and the other agent to represent the model for the dual task, then ask them to teach each other through a reinforcement learning process. Based on the feedback signals generated during this process (e.g., the language-model likelihood of the output of a model, and the reconstruction error of the original sentence after the primal and dual translations), we can iteratively update the two models until convergence (e.g., using the policy gradient methods). We call the corresponding approach to neural machine translation \emph{dual-NMT}. Experiments show that dual-NMT works very well on English$\leftrightarrow$French translation; especially, by learning from monolingual data (with 10% bilingual data for warm start), it achieves a comparable accuracy to NMT trained from the full bilingual data for the French-to-English translation task.

References in corpus (3)

• ADADELTA: An Adaptive Learning Rate Method
• Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
• On Using Monolingual Corpora in Neural Machine Translation

Cited by in corpus (34)

• CoaCor: Code Annotation for Code Retrieval with Reinforcement Learning
• The Missing Ingredient in Zero-Shot Neural Machine Translation
• Soft Contextual Data Augmentation for Neural Machine Translation
• A Dual Reinforcement Learning Framework for Unsupervised Text Style Transfer
• Robust Neural Machine Translation with Doubly Adversarial Inputs
• DDTCDR: Deep Dual Transfer Cross Domain Recommendation
• Corpus Augmentation by Sentence Segmentation for Low-Resource Neural Machine Translation
• Revisiting Low-Resource Neural Machine Translation: A Case Study
• WaveCycleGAN2: Time-domain Neural Post-filter for Speech Waveform Generation
• Estimating Gradients for Discrete Random Variables by Sampling without Replacement
• Synthetic QA Corpora Generation with Roundtrip Consistency
• Retrieving Sequential Information for Non-Autoregressive Neural Machine Translation
• Logician: A Unified End-to-End Neural Approach for Open-Domain Information Extraction
• DAL: Dual Adversarial Learning for Dialogue Generation
• Information Maximizing Visual Question Generation
• Reconstruct and Represent Video Contents for Captioning via Reinforcement Learning
• Extract and Edit: An Alternative to Back-Translation for Unsupervised Neural Machine Translation
• Jointly Learning Semantic Parser and Natural Language Generator via Dual Information Maximization
• Multi-task Learning for Multilingual Neural Machine Translation
• Deformation-aware Unpaired Image Translation for Pose Estimation on Laboratory Animals
• Mimicking the In-Camera Color Pipeline for Camera-Aware Object Compositing
• Semantic Parsing with Dual Learning
• Using Monolingual Data in Neural Machine Translation: a Systematic Study
• Hierarchical Photo-Scene Encoder for Album Storytelling
• Grounded and Controllable Image Completion by Incorporating Lexical Semantics
• Iterative Domain-Repaired Back-Translation
• Semi-supervised Text Style Transfer: Cross Projection in Latent Space
• Depth Growing for Neural Machine Translation
• Transformation of Dense and Sparse Text Representations
• Cycle-SUM: Cycle-consistent Adversarial LSTM Networks for Unsupervised Video Summarization
• Unsupervised Domain Adaptive Object Detection using Forward-Backward Cyclic Adaptation
• Improving Multi-Turn Response Selection Models with Complementary Last-Utterance Selection by Instance Weighting
• Learning Dual Retrieval Module for Semi-supervised Relation Extraction
• Iterative Dual Domain Adaptation for Neural Machine Translation