Maximum-Likelihood Augmented Discrete Generative Adversarial Networks

Abstract

Despite the successes in capturing continuous distributions, the application of generative adversarial networks (GANs) to discrete settings, like natural language tasks, is rather restricted. The fundamental reason is the difficulty of back-propagation through discrete random variables combined with the inherent instability of the GAN training objective. To address these problems, we propose Maximum-Likelihood Augmented Discrete Generative Adversarial Networks. Instead of directly optimizing the GAN objective, we derive a novel and low-variance objective using the discriminator's output that follows corresponds to the log-likelihood. Compared with the original, the new objective is proved to be consistent in theory and beneficial in practice. The experimental results on various discrete datasets demonstrate the effectiveness of the proposed approach.

11 pages, 3 figures

References in corpus (12)

• Conditional Generative Adversarial Nets
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Improved Techniques for Training GANs
• On the Properties of Neural Machine Translation: Encoder-Decoder Approaches
• Professor Forcing: A New Algorithm for Training Recurrent Networks
• Mode Regularized Generative Adversarial Networks
• Adversarial Learning for Neural Dialogue Generation
• Generative Visual Manipulation on the Natural Image Manifold
• Boundary-Seeking Generative Adversarial Networks
• Reward Augmented Maximum Likelihood for Neural Structured Prediction
• Temporal Generative Adversarial Nets with Singular Value Clipping
• Variational Neural Machine Translation

Cited by in corpus (66)

• Improved Training of Wasserstein GANs
• Deep learning for molecular design - a review of the state of the art
• Style Transfer from Non-Parallel Text by Cross-Alignment
• Objective-Reinforced Generative Adversarial Networks (ORGAN) for Sequence Generation Models
• How Generative Adversarial Networks and Their Variants Work: An Overview
• MaskGAN: Better Text Generation via Filling in the______
• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• Synthesizing Tabular Data using Generative Adversarial Networks
• NetGAN: Generating Graphs via Random Walks
• Long Text Generation via Adversarial Training with Leaked Information
• Texygen: A Benchmarking Platform for Text Generation Models
• Survey on reinforcement learning for language processing
• Generating and designing DNA with deep generative models
• Adversarial Generation of Natural Language
• Language Generation with Recurrent Generative Adversarial Networks without Pre-training
• Language GANs Falling Short
• Neural Text Generation: Past, Present and Beyond
• Unsupervised Text Style Transfer using Language Models as Discriminators
• Defense Methods Against Adversarial Examples for Recurrent Neural Networks
• Adversarially Regularized Autoencoders
• Unsupervised Cipher Cracking Using Discrete GANs
• Neural Language Generation: Formulation, Methods, and Evaluation
• CoT: Cooperative Training for Generative Modeling of Discrete Data
• Justifying Diagnosis Decisions by Deep Neural Networks
• A Review of Learning with Deep Generative Models from Perspective of Graphical Modeling
• SALSA-TEXT : self attentive latent space based adversarial text generation
• Pulsar Candidate Identification with Artificial Intelligence Techniques
• Correlated discrete data generation using adversarial training
• Jointly Measuring Diversity and Quality in Text Generation Models
• Self-Adversarial Learning with Comparative Discrimination for Text Generation
• Improving GAN Training with Probability Ratio Clipping and Sample Reweighting
• Weakly-supervised Knowledge Graph Alignment with Adversarial Learning
• Improving Conditional Sequence Generative Adversarial Networks by Stepwise Evaluation
• ColdGANs: Taming Language GANs with Cautious Sampling Strategies
• ARAML: A Stable Adversarial Training Framework for Text Generation
• Reward Constrained Interactive Recommendation with Natural Language Feedback
• DLGNet: A Transformer-based Model for Dialogue Response Generation
• Judge the Judges: A Large-Scale Evaluation Study of Neural Language Models for Online Review Generation
• Garment Design with Generative Adversarial Networks
• Revision in Continuous Space: Unsupervised Text Style Transfer without Adversarial Learning
• TextBox: A Unified, Modularized, and Extensible Framework for Text Generation
• Adding A Filter Based on The Discriminator to Improve Unconditional Text Generation
• Unsupervised Text Embedding Space Generation Using Generative Adversarial Networks for Text Synthesis
• Text Generation Based on Generative Adversarial Nets with Latent Variable
• Using GAN-based models to sentimental analysis on imbalanced datasets in education domain
• Collaborative Training of GANs in Continuous and Discrete Spaces for Text Generation
• Improving Text Generation with Student-Forcing Optimal Transport
• Meta-CoTGAN: A Meta Cooperative Training Paradigm for Improving Adversarial Text Generation
• Adversarial Bootstrapping for Dialogue Model Training
• Semi-Implicit Generative Model
• Sample weighting as an explanation for mode collapse in generative adversarial networks
• Adversarial Sub-sequence for Text Generation
• SparseGAN: Sparse Generative Adversarial Network for Text Generation
• To Beam Or Not To Beam: That is a Question of Cooperation for Language GANs
• Classification of sparsely labeled spatio-temporal data through semi-supervised adversarial learning
• Adversarial Training of Word2Vec for Basket Completion
• ReGAN: RE[LAX|BAR|INFORCE] based Sequence Generation using GANs
• TaylorGAN: Neighbor-Augmented Policy Update for Sample-Efficient Natural Language Generation
• Generating Symbolic Reasoning Problems with Transformer GANs
• Evaluating Computational Language Models with Scaling Properties of Natural Language
• The Detection of Distributional Discrepancy for Text Generation
• Generation of Synthetic Electronic Medical Record Text
• Generative Adversarial Networks for Annotated Data Augmentation in Data Sparse NLU
• A Novel Approach to Curiosity and Explainable Reinforcement Learning via Interpretable Sub-Goals
• Improving Adversarial Text Generation by Modeling the Distant Future
• Adversarially-Trained Normalized Noisy-Feature Auto-Encoder for Text Generation