Discriminative Regularization for Generative Models

Abstract

We explore the question of whether the representations learned by classifiers can be used to enhance the quality of generative models. Our conjecture is that labels correspond to characteristics of natural data which are most salient to humans: identity in faces, objects in images, and utterances in speech. We propose to take advantage of this by using the representations from discriminative classifiers to augment the objective function corresponding to a generative model. In particular we enhance the objective function of the variational autoencoder, a popular generative model, with a discriminative regularization term. We show that enhancing the objective function in this way leads to samples that are clearer and have higher visual quality than the samples from the standard variational autoencoders.

References in corpus (4)

• Deep Learning in Neural Networks: An Overview
• Stochastic Backpropagation and Approximate Inference in Deep Generative Models
• Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
• DRAW: A Recurrent Neural Network For Image Generation

Cited by in corpus (17)

• Adversarially Learned Inference
• Neural Photo Editing with Introspective Adversarial Networks
• Learning a Driving Simulator
• Towards a Neural Statistician
• Towards Deeper Understanding of Variational Autoencoding Models
• Super-Resolution with Deep Convolutional Sufficient Statistics
• Stacked Generative Adversarial Networks
• Transformation-Grounded Image Generation Network for Novel 3D View Synthesis
• Hierarchical Adversarially Learned Inference
• Light Field Synthesis by Training Deep Network in the Refocused Image Domain
• Deep Feature Consistent Variational Autoencoder
• Improving Sampling from Generative Autoencoders with Markov Chains
• Improving Consistency and Correctness of Sequence Inpainting using Semantically Guided Generative Adversarial Network
• Hybrid VAE: Improving Deep Generative Models using Partial Observations
• Achieving Explainability for Plant Disease Classification with Disentangled Variational Autoencoders
• Sound2Sight: Generating Visual Dynamics from Sound and Context
• Melody-Conditioned Lyrics Generation with SeqGANs