Generative Adversarial Networks: An Overview

By Antonia Creswell, Tom White, Vincent Dumoulin, Kai Arulkumaran, Biswa Sengupta, Anil Anthony Bharath. 2017

3,805 citations recorded by the source corpus.

Generative adversarial networks (GANs) provide a way to learn deep representations without extensively annotated training data. They achieve this by deriving backpropagation signals through a competitive process involving a pair of networks. The representations that can be learned by GANs may be used in a variety of applications, including image synthesis, semantic image editing, style transfer, image superresolution, and classification. The aim of this review article is to provide an overview of GANs for the signal processing community, drawing on familiar analogies and concepts where possible. In addition to identifying different methods for training and constructing GANs, we also point to remaining challenges in their theory and application.

Subfields: Computer Vision and Pattern Recognition, Generative Models

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Generative Models

Citations in the corpus

CLIP-GEN: Language-Free Training of a Text-to-Image Generator with CLIP (2022, background): “[12] introduces self-attention layers [36] to the bottleneck of the convolutional architecture with the expect to capture longrange interactions in high-resolution images, and adversarial training losses [10] to enforce the learning of perceptually rich codebooks.”
An overview of deep learning in medical imaging focusing on MRI (2018, background): “ have seen impressive results for photo-realistic image synthesis using deep learning techniques, especially generative adversarial networks (GANs, introduced by Goodfellow et al. in 2014 [80]), e.g. [205, 206, 207]. These can also be used for biological image synthesis [208, 209] and text-to-image ”
How to See Hidden Patterns in Metamaterials with Interpretable Machine Learning (2021, methodology): “Here, we did not compare with deep generative models such as GANs.”
ProDiff: Progressive Fast Diffusion Model for High-Quality Text-to-Speech (2022, methodology): “As shown in Table 2, we can see that ProDiff achieves higher scores than several one-shot methods, while the GAN-based method generates samples with minimal diversity, which is expected for the following reasons: 1) It is well-known that mode collapse [3] easily appears in the strongly conditional g”
Editing in Style: Uncovering the Local Semantics of GANs (2020, background): “For a thorough review of the GAN literature we refer the reader to recent surveys in [7, 13, 27].”
A Survey on Generative Adversarial Networks: Variants, Applications, and Training (2020, background): “ surveys and significance In literature, several other relevant surveys (e.g., Gui et al. 2020 [2]; Wang et al. 2019 [3]; PAN et al. 2019 [4]; Hong et al. 2019 [5]; Hitawala. 2018 [6]; Creswell. 2018 [7]; Vuppuluri et al. 2017 [8]) of GAN have studied to investigate the recent trends, and their pote”
MedGAN: Medical Image Translation using GANs (2018, background): “nt performance in the generation of photo-realistic images, GAN has evoked much interest in the computer vision community. A large body of literature tracks the recent theoretical advancement of GANs [23], [24]. Alex Radford [25] made the natural extension of GANs from fully connected networks to CN”
Synthesis of Realistic ECG using Generative Adversarial Networks (2019, background): “In addition, there is a lack of consensus in current research on the best choice of evaluation metric for GANs [11].”
A Brief Overview of ChatGPT: The History, Status Quo and Potential Future Development (2023, methodology): “With the iterative development of AI algorithms and network structures [4], [5], significant progress has been made in AIGC.”
Brain Imaging Generation with Latent Diffusion Models (2022, background): “Nowadays, Generative Adversarial Networks (GANs) have been applied in various ﬁelds to create synthetic images, producing realistic and clear images and achieving impressive performance [7,32].”
Generative artificial intelligence: a systematic review and applications (2024, neutral): “Moreover, the architecture that it affords is quite unique from the conventional Deep Neural Networks [27].”
An Introduction to Image Synthesis with Generative Adversarial Nets (2018, background): “Also, there is a recent review paper on GAN [15] which reviews several GAN architectures and training techniques, and introduces some applications of GAN.”