Adversarial Text Generation Without Reinforcement Learning

10/11/2018
by   David Donahue, et al.
0

Generative Adversarial Networks (GANs) have experienced a recent surge in popularity, performing competitively in a variety of tasks, especially in computer vision. However, GAN training has shown limited success in natural language processing. This is largely because sequences of text are discrete, and thus gradients cannot propagate from the discriminator to the generator. Recent solutions use reinforcement learning to propagate approximate gradients to the generator, but this is inefficient to train. We propose to utilize an autoencoder to learn a low-dimensional representation of sentences. A GAN is then trained to generate its own vectors in this space, which decode to realistic utterances. We report both random and interpolated samples from the generator. Visualization of sentence vectors indicate our model correctly learns the latent space of the autoencoder. Both human ratings and BLEU scores show that our model generates realistic text against competitive baselines.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
07/18/2017

Optimizing the Latent Space of Generative Networks

Generative Adversarial Networks (GANs) have been shown to be able to sam...
research
04/23/2019

TextKD-GAN: Text Generation using KnowledgeDistillation and Generative Adversarial Networks

Text generation is of particular interest in many NLP applications such ...
research
11/13/2017

ACtuAL: Actor-Critic Under Adversarial Learning

Generative Adversarial Networks (GANs) are a powerful framework for deep...
research
06/12/2020

Improving GAN Training with Probability Ratio Clipping and Sample Reweighting

Despite success on a wide range of problems related to vision, generativ...
research
06/12/2017

Adversarial Feature Matching for Text Generation

The Generative Adversarial Network (GAN) has achieved great success in g...
research
02/23/2023

Improved Training of Mixture-of-Experts Language GANs

Despite the dramatic success in image generation, Generative Adversarial...
research
09/01/2021

OptAGAN: Entropy-based finetuning on text VAE-GAN

Transfer learning through large pre-trained models has changed the lands...

Please sign up or login with your details

Forgot password? Click here to reset