Improving Adversarial Text Generation by Modeling the Distant Future

Abstract

Auto-regressive text generation models usually focus on local fluency, and may cause inconsistent semantic meaning in long text generation. Further, automatically generating words with similar semantics is challenging, and hand-crafted linguistic rules are difficult to apply. We consider a text planning scheme and present a model-based imitation-learning approach to alleviate the aforementioned issues. Specifically, we propose a novel guider network to focus on the generative process over a longer horizon, which can assist next-word prediction and provide intermediate rewards for generator optimization. Extensive experiments demonstrate that the proposed method leads to improved performance.

arXiv commentACL 2020. arXiv admin note: substantial text overlap with arXiv:1811.00696

References in corpus (17)

• Adam: A Method for Stochastic Optimization
• A Deep Reinforced Model for Abstractive Summarization
• Model-Ensemble Trust-Region Policy Optimization
• An Actor-Critic Algorithm for Sequence Prediction
• GANS for Sequences of Discrete Elements with the Gumbel-softmax Distribution
• Maximum-Likelihood Augmented Discrete Generative Adversarial Networks
• Long Text Generation via Adversarial Training with Leaked Information
• Language GANs Falling Short
• Adversarial Text Generation via Feature-Mover's Distance
• Deep Reinforcement Learning-based Image Captioning with Embedding Reward
• Twin Networks: Matching the Future for Sequence Generation
• Unsupervised Text Style Transfer using Language Models as Discriminators
• Actor-Critic based Training Framework for Abstractive Summarization
• Data Generation as Sequential Decision Making
• Model-based Adversarial Imitation Learning
• Policy Optimization as Wasserstein Gradient Flows
• Improving Sequence-to-Sequence Learning via Optimal Transport