# Triple Generative Adversarial Networks

arXiv: https://arxiv.org/abs/1912.09784 · HTML: https://mlsanity.com/paper/1912.09784/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Dec 2019.

## Built on by (most cited first, 11 of 11)

- [ORDisCo: Effective and Efficient Usage of Incremental Unlabeled Data for Semi-supervised Continual Learning](https://mlsanity.com/paper/2101.00407/) (2021)
- [Diffusion Models and Semi-Supervised Learners Benefit Mutually with Few Labels](https://mlsanity.com/paper/2302.10586/) (2023)
- [AdvMIL: Adversarial Multiple Instance Learning for the Survival Analysis on Whole-Slide Images](https://mlsanity.com/paper/2212.06515/) (2022)
- [Erasing Concepts, Steering Generations: A Comprehensive Survey of Concept Suppression](https://mlsanity.com/paper/2505.19398/) (2025)
- [Provable Unrestricted Adversarial Training without Compromise with Generalizability](https://mlsanity.com/paper/2301.09069/) (2023)
- [Deep Generative Modeling on Limited Data with Regularization by Nontransferable Pre-trained Models](https://mlsanity.com/paper/2208.14133/) (2022)
- [Global Nash Equilibrium in Non-convex Multi-player Game: Theory and Algorithms](https://mlsanity.com/paper/2301.08015/) (2023)
- [Parameters, Properties, and Process: Conditional Neural Generation of Realistic SEM Imagery Towards ML-assisted Advanced Manufacturing](https://mlsanity.com/paper/2302.08495/) (2023)
- [On the link between generative semi-supervised learning and generative open-set recognition](https://mlsanity.com/paper/2303.11702/) (2023)
- [Semi-Supervised Bayesian GANs with Log-Signatures for Uncertainty-Aware Credit Card Fraud Detection](https://mlsanity.com/paper/2509.00931/) (2025)
- [SPARSE Data, Rich Results: Few-Shot Semi-Supervised Learning via Class-Conditioned Image Translation](https://mlsanity.com/paper/2508.06429/) (2025)

## Abstract (arXiv)

We propose a unified game-theoretical framework to perform classification and conditional image generation given limited supervision. It is formulated as a three-player minimax game consisting of a generator, a classifier and a discriminator, and therefore is referred to as Triple Generative Adversarial Network (Triple-GAN). The generator and the classifier characterize the conditional distributions between images and labels to perform conditional generation and classification, respectively. The discriminator solely focuses on identifying fake image-label pairs. Under a nonparametric assumption, we prove the unique equilibrium of the game is that the distributions characterized by the generator and the classifier converge to the data distribution. As a byproduct of the three-player mechanism, Triple-GAN is flexible to incorporate different semi-supervised classifiers and GAN architectures. We evaluate Triple-GAN in two challenging settings, namely, semi-supervised learning and the extreme low data regime. In both settings, Triple-GAN can achieve excellent classification results and generate meaningful samples in a specific class simultaneously. In particular, using a commonly adopted 13-layer CNN classifier, Triple-GAN outperforms extensive semi-supervised learning methods substantially on more than 10 benchmarks no matter data augmentation is applied or not.