# On Catastrophic Forgetting and Mode Collapse in Generative Adversarial Networks

arXiv: https://arxiv.org/abs/1807.04015 · HTML: https://mlsanity.com/paper/1807.04015/

## Key facts

- Cited by 17 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 11 Jul 2018.

## Built on by (most cited first, 17 of 17)

- [A Comprehensive Survey of Forgetting in Deep Learning Beyond Continual Learning](https://mlsanity.com/paper/2307.09218/) (2023)
- [Diffusion-Based Representation Learning](https://mlsanity.com/paper/2105.14257/) (2021)
- [EHRDiff: Exploring Realistic EHR Synthesis with Diffusion Models](https://mlsanity.com/paper/2303.05656/) (2023)
- [Learning C to x86 Translation: An Experiment in Neural Compilation](https://mlsanity.com/paper/2108.07639/) (2021)
- [RFWave: Multi-band Rectified Flow for Audio Waveform Reconstruction](https://mlsanity.com/paper/2403.05010/) (2024)
- [Evaluating the Performance and Robustness of LLMs in Materials Science Q&A and Property Predictions](https://mlsanity.com/paper/2409.14572/) (2024)
- [Instability and Local Minima in GAN Training with Kernel Discriminators](https://mlsanity.com/paper/2208.09938/) (2022)
- [Guided Discrete Diffusion for Electronic Health Record Generation](https://mlsanity.com/paper/2404.12314/) (2024)
- [Image-to-Height Domain Translation for Synthetic Aperture Sonar](https://mlsanity.com/paper/2112.06307/) (2021)
- [Novel Applications for VAE-based Anomaly Detection Systems](https://mlsanity.com/paper/2204.12577/) (2022)
- [Towards Robustness and Diversity: Continual Learning in Dialog Generation with Text-Mixup and Batch Nuclear-Norm Maximization](https://mlsanity.com/paper/2403.10894/) (2024)
- [Domain Translation via Latent Space Mapping](https://mlsanity.com/paper/2212.03361/) (2022)
- [FREDSR: Fourier Residual Efficient Diffusive GAN for Single Image Super Resolution](https://mlsanity.com/paper/2211.16678/) (2022)
- [Black-Box Training Data Identification in GANs via Detector Networks](https://mlsanity.com/paper/2310.12063/) (2023)
- [Statistical Mechanics Calculations Using Variational Autoregressive Networks and Quantum Annealing](https://mlsanity.com/paper/2404.19274/) (2024)
- [HUydra: Full-Range Lung CT Synthesis via Multiple HU Interval Generative Modelling](https://mlsanity.com/paper/2603.23041/) (2026)
- [A Wasserstein GAN-based climate scenario generator for risk management and insurance: the case of soil subsidence](https://mlsanity.com/paper/2605.06678/) (2026)

## Abstract (arXiv)

In this paper, we show that Generative Adversarial Networks (GANs) suffer from catastrophic forgetting even when they are trained to approximate a single target distribution. We show that GAN training is a continual learning problem in which the sequence of changing model distributions is the sequence of tasks to the discriminator. The level of mismatch between tasks in the sequence determines the level of forgetting. Catastrophic forgetting is interrelated to mode collapse and can make the training of GANs non-convergent. We investigate the landscape of the discriminator's output in different variants of GANs and find that when a GAN converges to a good equilibrium, real training datapoints are wide local maxima of the discriminator. We empirically show the relationship between the sharpness of local maxima and mode collapse and generalization in GANs. We show how catastrophic forgetting prevents the discriminator from making real datapoints local maxima, and thus causes non-convergence. Finally, we study methods for preventing catastrophic forgetting in GANs.