Abstract:Generative Adversarial Networks (GANs) are an unsupervised generative model that learns data distribution through adversarial training. However, recent experiments indicated that GANs are difficult to train due to the requirement of optimization in the high dimensional parameter space and the zero gradient problem. In this work, we propose a Self Sparse Generative Adversarial Network (Self-Sparse GAN) that reduces the parameter space and alleviates the zero gradient problem. In the Self-Sparse GAN, we design a Self-Adaptive Sparse Transform Module (SASTM) comprising the sparsity decomposition and feature-map recombination, which can be applied on multi-channel feature maps to obtain sparse feature maps. The key idea of Self-Sparse GAN is to add the SASTM following every deconvolution layer in the generator, which can adaptively reduce the parameter space by utilizing the sparsity in multi-channel feature maps. We theoretically prove that the SASTM can not only reduce the search space of the convolution kernel weight of the generator but also alleviate the zero gradient problem by maintaining meaningful features in the Batch Normalization layer and driving the weight of deconvolution layers away from being negative. The experimental results show that our method achieves the best FID scores for image generation compared with WGAN-GP on MNIST, Fashion-MNIST, CIFAR-10, STL-10, mini-ImageNet, CELEBA-HQ, and LSUN bedrooms, and the relative decrease of FID is 4.76% ~ 21.84%.

Lightweight Generative Adversarial Networks Based on Ghost Module.

Generative-Adversarial-Networks-based Ghost Recognition

Training Generative Adversarial Networks in One Stage

UGC: Unified GAN Compression for Efficient Image-to-Image Translation

Self Sparse Generative Adversarial Networks

A lightweight ensemble discriminator for Generative Adversarial Networks

Latent Dirichlet Allocation Based Generative Adversarial Networks.

Ghost-Unet: multi-stage network for image deblurring via lightweight subnet learning

RepGhost: A Hardware-Efficient Ghost Module via Re-parameterization

Underwater Ghost Imaging Based on Generative Adversarial Networks with High Imaging Quality.

Lightweight Cartoonlization Method Based on Generative Adversarial Network

Lightweight Semantic Image Synthesis with Mutable Framework

L-GhostNet: Extract Better Quality Features

Optimized Pointwise Convolution Operation by Ghost Blocks

Lightweight dual-path octave generative adversarial networks for few-shot image generation

Image-to-Image Style Transfer Based on the Ghost Module

GhostNet: More Features from Cheap Operations

Lightweight Privacy-Preserving GAN Framework for Model Training and Image Synthesis

Densely Stacked Generative Adversarial Networks

Ghost-Depth: A Lightweight Encoder-Decoder Network for Monocular Depth Estimation.

Ghost shuffle lightweight pose network with effective feature representation and learning for human pose estimation