Self-Attention Generative Adversarial Networks (2018-05-21T00:00:00.000000Z)

TL;DR

The proposed SAGAN achieves the state-of-the-art results, boosting the best published Inception score from 36.8 to 52.52 and reducing Frechet Inception distance from 27.62 to 18.65 on the challenging ImageNet dataset.

Abstract

In this paper, we propose the Self-Attention Generative Adversarial Network (SAGAN) which allows attention-driven, long-range dependency modeling for image generation tasks. Traditional convolutional GANs generate high-resolution details as a function of only spatially local points in lower-resolution feature maps. In SAGAN, details can be generated using cues from all feature locations. Moreover, the discriminator can check that highly detailed features in distant portions of the image are consistent with each other. Furthermore, recent work has shown that generator conditioning affects GAN performance. Leveraging this insight, we apply spectral normalization to the GAN generator and find that this improves training dynamics. The proposed SAGAN achieves the state-of-the-art results, boosting the best published Inception score from 36.8 to 52.52 and reducing Frechet Inception distance from 27.62 to 18.65 on the challenging ImageNet dataset. Visualization of the attention layers shows that the generator leverages neighborhoods that correspond to object shapes rather than local regions of fixed shape.

Authors

I. Goodfellow

19 papers

Dimitris N. Metaxas

9 papers

Augustus Odena

4 papers

TL;DR

Abstract

Authors

References54 items

Open Questions about Generative Adversarial Networks

HYPE: Human eYe Perceptual Evaluation of Generative Models

Semantic Image Synthesis With Spatially-Adaptive Normalization

A Style-Based Generator Architecture for Generative Adversarial Networks

Large Scale GAN Training for High Fidelity Natural Image Synthesis

Discriminator Rejection Sampling

Skill Rating for Generative Models

The relativistic discriminator: a key element missing from standard GAN

Is Generator Conditioning Causally Related to GAN Performance?

cGANs with Projection Discriminator

Spectral Normalization for Generative Adversarial Networks

Improving GANs Using Optimal Transport

Image Transformer

Geometry Score: A Method For Comparing Generative Adversarial Networks

Inferring Semantic Layout for Hierarchical Text-to-Image Synthesis

PixelSNAIL: An Improved Autoregressive Generative Model

AttnGAN: Fine-Grained Text to Image Generation with Attentional Generative Adversarial Networks

Non-local Neural Networks

Progressive Growing of GANs for Improved Quality, Stability, and Variation

StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks

GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium

Attention is All you Need

SegAN: Adversarial Network with Multi-scale L1 Loss for Medical Image Segmentation

Geometric GAN

Improved Training of Wasserstein GANs

Deep and Hierarchical Implicit Models

Hierarchical Implicit Models and Likelihood-Free Variational Inference

Wasserstein GAN

StackGAN: Text to Photo-Realistic Image Synthesis with Stacked Generative Adversarial Networks

Image-to-Image Translation with Conditional Adversarial Networks

Mode Regularized Generative Adversarial Networks

Unrolled Generative Adversarial Networks

Unsupervised Cross-Domain Image Generation

Conditional Image Synthesis with Auxiliary Classifier GANs

Amortised MAP Inference for Image Super-resolution

Learning What and Where to Draw

Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network

Energy-based Generative Adversarial Network

Coupled Generative Adversarial Networks

Conditional Image Generation with PixelCNN Decoders

Improved Techniques for Training GANs

A Decomposable Attention Model for Natural Language Inference

Generative Adversarial Text to Image Synthesis

Long Short-Term Memory-Networks for Machine Reading

Faster Asynchronous SGD

Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

Stacked Attention Networks for Image Question Answering

DRAW: A Recurrent Neural Network For Image Generation

Show, Attend and Tell: Neural Image Caption Generation with Visual Attention

Adam: A Method for Stochastic Optimization

Neural Machine Translation by Jointly Learning to Align and Translate

ImageNet Large Scale Visual Recognition Challenge

Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

GENERATIVE ADVERSARIAL NETS

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names