Information-Theoretic GAN Compression with Variational Energy-based Model

TL;DR

This paper introduces an information-theoretic method for compressing GANs by maximizing mutual information between teacher and student networks using a variational energy-based model, improving efficiency while maintaining quality.

Contribution

It presents a novel variational energy-based approach for GAN compression that is flexible and can be integrated into various generative models, enhancing compression performance.

Findings

Achieves superior GAN compression performance across multiple models

Effectively handles high-dimensional images with spatial dependencies

Compatible with diverse generative and dense prediction networks

Abstract

We propose an information-theoretic knowledge distillation approach for the compression of generative adversarial networks, which aims to maximize the mutual information between teacher and student networks via a variational optimization based on an energy-based model. Because the direct computation of the mutual information in continuous domains is intractable, our approach alternatively optimizes the student network by maximizing the variational lower bound of the mutual information. To achieve a tight lower bound, we introduce an energy-based model relying on a deep neural network to represent a flexible variational distribution that deals with high-dimensional images and consider spatial dependencies between pixels, effectively. Since the proposed method is a generic optimization algorithm, it can be conveniently incorporated into arbitrary generative adversarial networks and even dense prediction networks, e.g., image enhancement models. We demonstrate that the proposed algorithm achieves outstanding performance in model compression of generative adversarial networks consistently when combined with several existing models.