A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications

By Jie Gui, Zhenan Sun, Yonggang Wen, Dacheng Tao, Jieping Ye. IEEE Transactions on Knowledge and Data Engineering, 2020

1,136 citations recorded by the source corpus.

Generative adversarial networks (GANs) have recently become a hot research topic; however, they have been studied since 2014, and a large number of algorithms have been proposed. Nevertheless, few comprehensive studies explain the connections among different GAN variants and how they have evolved. In this paper, we attempt to provide a review of the various GAN methods from the perspectives of algorithms, theory, and applications. First, the motivations, mathematical representations, and structures of most GAN algorithms are introduced in detail, and we compare their commonalities and differences. Second, theoretical issues related to GANs are investigated. Finally, typical applications of GANs in image processing and computer vision, natural language processing, music, speech and audio, the medical field, and data science are discussed.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/fbc5486a1ffb9039dbb5046b84f0eb32e4ce8eea

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

1. Dacheng Tao
2. Zhangyang Wang
3. Jing Zhang
4. Ming-Hsuan Yang
5. Shiyu Chang
6. Xi Li
7. Luc Van Gool
8. Zhengwei Wang

Citations in the corpus

• Retrieval-Augmented Generation for AI-Generated Content: A Survey (2024, neutral): “1) Query-based RAG : Query-based RAG, originated from the idea of prompt augmentation, integrates the user’s query with insights from information fetched during the retrieval process, directly into the initial stage of the language model’s input.”
• Diffusion Probabilistic Modeling for Video Generation (2022, background): “Another line of sequential models relies on GANs [16,17,40,63–65], which—at inference time—can be either deterministic or stochastic.”
• Diffusion Models: A Comprehensive Survey of Methods and Applications (2022, methodology): “Generative Adversarial Networks (GANs) [40, 73, 84] mainly consist of two models: a generatorG and a discriminator D .”
• TransGAN: Two Transformers Can Make One Strong GAN (2021, background): “GANs (Gui et al., 2020) can be generalized to minimizing a large family of divergences, and are practically formulated as minimax optimization.”
• Generative Adversarial Networks (GANs) (2020, methodology): “VAE-GAN [72], AAE [73], AVB [74], ASVAE [75], MDGAN [40], Dist-GAN [76], α-GAN [25] Latent space autoencoder (S 13 (ii)) ALI [77], BiGAN [78], DALI [79], CV-BiGAN [80], MV-BiGAN [80], HALI [81], AGE [82], VEEGAN [83], MGGAN [84] Improved D (S 14 ) EBGAN [85], BEGAN [86], MAGAN [87], [88],…”
• Generative Adversarial Networks in Time Series: A Systematic Literature Review (2022, background): “The authors in [40] breakdown GANs into their constituent parts.”
• Deep Generative Modelling: A Comparative Review of VAEs, GANs, Normalizing Flows, Energy-Based and Autoregressive Models (2021, neutral): “[175], generative adversarial networks [70], [249], and energy-based models [202], however, naturally these dive into the intricacies of their respective method rather than”
• A Survey on Generative Adversarial Networks: Variants, Applications, and Training (2020, background): “A Review on Generative Adversarial Nets: Algorithms, Theory, and Applications [2] arXiv 2020”
• A Survey on Self-Supervised Learning: Algorithms, Applications, and Future Trends (2023, background): “Thiscategorizationincludesgenerativeadversarialnetworks (GANs)[35],therebypositioningthemwithintherealmofSSL.”
• TransGAN: Two Pure Transformers Can Make One Strong GAN, and That Can Scale Up (2021, methodology): “After its origin, GANs quickly embraced fully convolutional backbones [26], and inherited most successful designs from CNNs such as batch normalization, pooling, (Leaky) ReLU and more [38, 39, 40, 18].”
• Generative artificial intelligence (2023, positive): “…for instance, in image generation and manipulation, object detection and segmentation, and natural language processing (Aggarwal et al., 2021; Gui et al., 2023) Variational autoencoder (VAE) V ariational autoencoders employ a neural network to learn encoding compressed input data into a…”