Improving Neural Machine Translation with Conditional Sequence Generative Adversarial Nets

By Zhen-Le Yang, Feng Wang, Bo Xu. 2017

173 citations recorded by the source corpus.

This paper proposes an approach for applying GANs to NMT. We build a conditional sequence generative adversarial net which comprises of two adversarial sub models, a generator and a discriminator. The generator aims to generate sentences which are hard to be discriminated from human-translated sentences ( i.e., the golden target sentences); And the discriminator makes efforts to discriminate the machine-generated sentences from human-translated ones. The two sub models play a mini-max game and achieve the win-win situation when they reach a Nash Equilibrium. Additionally, the static sentence-level BLEU is utilized as the reinforced objective for the generator, which biases the generation towards high BLEU points. During training, both the dynamic discriminator and the static BLEU objective are employed to evaluate the generated sentences and feedback the evaluations to guide the learning of the generator. Experimental results show that the proposed model consistently outperforms the traditional RNNSearch and the newly emerged state-of-the-art Transformer on English-German and Chinese-English translation tasks.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Natural Language Processing, Generative Models

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Natural Language Processing

In Generative Models

Citations in the corpus

Improved Training of Wasserstein GANs (2017, methodology): “Other attempts at language modeling with GANs [31, 14, 29, 5, 15, 10] typically use discrete models and gradient estimators [27, 12, 16].”
Generating Informative and Diverse Conversational Responses via Adversarial Information Maximization (2018, background): “Adversarial learning [6, 26] has been shown to be successful in dialog generation, translation, image captioning and a series of natural language generation tasks [7, 12, 27, 28, 29, 30, 31, 32, 33, 34].”
Long Text Generation via Adversarial Training with Leaked Information (2017, background): “Generating text that mimics human’s expression has been studied for poem generation (Zhang and Lapata 2014), image captioning (Vinyals et al. 2015), dialogue system (Li et al. 2017) machine translation (Yang et al. 2017).”
Real-valued (Medical) Time Series Generation with Recurrent Conditional GANs (2017, methodology): “Recently, Conditional GAN architectures have been also used in natural language processing, including translation [27] and dialogue generation [16], where none of them uses an RNN as the preferred choice for the discriminator and, as previously mentioned, a RL approach is used to train the models…”
Adversarial Ranking for Language Generation (2017, background): “Secondly, most of the existing GANs assume the output of the discriminator to be a binary predicate indicating whether the given sentence is written by human or machine [4, 16, 18, 34, 35].”
Neural Machine Translation with Adequacy-Oriented Learning (2018, unknown): “…such as CNN-based NMT models (Gehring et al. 2017) and improved T RANSFORMER models (Shaw, Uszkoreit, and Vaswani 2018; Shen et al. 2018), as well as combining with other advanced techniques in reinforcement learning and adversarial learning (Bahdanau et al. 2017; Yu et al. 2017; Yang et al. 2018).”
TextKD-GAN: Text Generation Using Knowledge Distillation and Generative Adversarial Networks (2019, unknown): “GANs have been exploited recently in various NLP applications such as machine translation [24,25], dialogue models [15], question answering [26], and natural language generation [7,20,19,13,28,29].”
Online Distilling from Checkpoints for Neural Machine Translation (2019, result): “As shown in Table 1, our baseline is comparable to the other two recent published results (Zhang et al. (2018b), Yang et al. (2018)).”
Unsupervised Neural Machine Translation with Weight Sharing (2018, methodology): “Since the machine translation is a sequence generation problem, following (Yang et al., 2017), we leverage policy gradient re-inforcement training to back-propagate the assessment.”
VIBE: Video Inference for Human Body Pose and Shape Estimation (2019, background): “Recent works have incorporated GANs into recurrent architectures to model sequence-to-sequence tasks like machine translation [59, 67, 68].”
On the Weaknesses of Reinforcement Learning for Neural Machine Translation (2019, methodology): “We turn to demonstrate that the initially most probable tokens will initially gain probability mass, even if they are not the most rewarding, yielding a P K E. Caccia et al. (2018) recently observed in the context of language modeling using GANs that performance gains similar to those GAN yield can ”
Neural Models of Text Normalization for Speech Applications (2019, background): “2014; Goodfellow 2016), which have achieved impressive results in vision-related tasks but which have been also applied to NLP tasks including machine translation (Wu et al. 2017; Yang et al. 2018).”
Neural Machine Translation for Low-resource Languages: A Survey (2021, methodology): “Most of the architectures for unsupervised NMT make use of the adversarial frameworks of Generative Adversarial Networks (GANs) and contain the following three steps [202]: i) initialization, ii) back-translation and iii) discriminative classiier.”
Modeling Fluency and Faithfulness for Diverse Neural Machine Translation (2019, methodology): “This series of work (Wu et al. 2018; Yang et al. 2018; Geng et al. 2018; Kreutzer, Uyheng, and Riezler 2018) samples a translation from all the possible translation and performs gradient descent through the translation with a sequence-level reward under the framework of reinforcement learning .”
Learning to Repair Software Vulnerabilities with Generative Adversarial Networks (2018, methodology): “This can easily be performed when paired data is available, by providing the discriminator with both x and y , thereby formulating the problem as in the conditional approach of Mirza and Osindero [25, 26].”
Conditional WaveGAN (2018, unknown): “Usage of generative models in natural language processing is also in progress with notabld ﬁelds including text modelling [28, 21], dialogue generation [16], and neural machine translation [27].”
Towards Robust Neural Machine Translation (2018, background): “Adversarial Learning Generative Adversarial Network (GAN) (Goodfellow et al., 2014) and its related derivative have been widely applied in computer vision (Radford et al., 2015; Sali-mans et al., 2016) and natural language processing (Li et al., 2017; Yang et al., 2018).”
Adversarial Learning of Task-Oriented Neural Dialog Models (2018, background): “Yang et al. (2017) proposed using adversarial learning in neural machine translation.”