Unsupervised Text Style Transfer using Language Models as Discriminators

Abstract

Binary classifiers are often employed as discriminators in GAN-based unsupervised style transfer systems to ensure that transferred sentences are similar to sentences in the target domain. One difficulty with this approach is that the error signal provided by the discriminator can be unstable and is sometimes insufficient to train the generator to produce fluent language. In this paper, we propose a new technique that uses a target domain language model as the discriminator, providing richer and more stable token-level feedback during the learning process. We train the generator to minimize the negative log likelihood (NLL) of generated sentences, evaluated by the language model. By using a continuous approximation of discrete sampling under the generator, our model can be trained using back-propagation in an end- to-end fashion. Moreover, our empirical results show that when using a language model as a structured discriminator, it is possible to forgo adversarial steps during training, making the process more stable. We compare our model with previous work using convolutional neural networks (CNNs) as discriminators and show that our approach leads to improved performance on three tasks: word substitution decipherment, sentiment modification, and related language translation.

NeurIPS camera ready

References in corpus (29)

• Adam: A Method for Stochastic Optimization
• Sequence to Sequence Learning with Neural Networks
• Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Categorical Reparameterization with Gumbel-Softmax
• Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
• Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
• A Neural Conversational Model
• Improved Techniques for Training GANs
• InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets
• Convolutional Neural Networks for Sentence Classification
• Energy-based Generative Adversarial Network
• Style Transfer from Non-Parallel Text by Cross-Alignment
• Dual Learning for Machine Translation
• On Using Monolingual Corpora in Neural Machine Translation
• Towards Principled Methods for Training Generative Adversarial Networks
• Professor Forcing: A New Algorithm for Training Recurrent Networks
• Adversarial Learning for Neural Dialogue Generation
• Unsupervised Machine Translation Using Monolingual Corpora Only
• Show and Tell: A Neural Image Caption Generator
• Maximum-Likelihood Augmented Discrete Generative Adversarial Networks
• A Network-based End-to-End Trainable Task-oriented Dialogue System
• Toward Controlled Generation of Text
• Improved Variational Autoencoders for Text Modeling using Dilated Convolutions
• Unsupervised Neural Machine Translation
• Delete, Retrieve, Generate: A Simple Approach to Sentiment and Style Transfer
• Style Transfer in Text: Exploration and Evaluation
• Unsupervised Cipher Cracking Using Discrete GANs
• Unsupervised Learning of Predictors from Unpaired Input-Output Samples

Cited by in corpus (19)

• Multiple-Attribute Text Style Transfer
• Adversarially Regularized Autoencoders
• Controllable Data Generation by Deep Learning: A Review
• Polyjuice: Generating Counterfactuals for Explaining, Evaluating, and Improving Models
• MALA: Cross-Domain Dialogue Generation with Action Learning
• "Mask and Infill" : Applying Masked Language Model to Sentiment Transfer
• Deep Extrapolation for Attribute-Enhanced Generation
• Connecting the Dots Between MLE and RL for Sequence Prediction
• From Theories on Styles to their Transfer in Text: Bridging the Gap with a Hierarchical Survey
• The Daunting Task of Real-World Textual Style Transfer Auto-Evaluation
• ER-AE: Differentially Private Text Generation for Authorship Anonymization
• Review of Text Style Transfer Based on Deep Learning
• Boosting Naturalness of Language in Task-oriented Dialogues via Adversarial Training
• Protecting Anonymous Speech: A Generative Adversarial Network Methodology for Removing Stylistic Indicators in Text
• A Text Reassembling Approach to Natural Language Generation
• TextSETTR: Few-Shot Text Style Extraction and Tunable Targeted Restyling
• TransSent: Towards Generation of Structured Sentences with Discourse Marker
• GTAE: Graph-Transformer based Auto-Encoders for Linguistic-Constrained Text Style Transfer
• Improving Adversarial Text Generation by Modeling the Distant Future