# Combating Mode Collapse in GAN training: An Empirical Analysis using Hessian Eigenvalues

arXiv: https://arxiv.org/abs/2012.09673 · HTML: https://mlsanity.com/paper/2012.09673/

## Key facts

- Cited by 27 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Dec 2020.

## Built on by (most cited first, 27 of 27)

- [DiffuseLoco: Real-Time Legged Locomotion Control with Diffusion from Offline Datasets](https://mlsanity.com/paper/2404.19264/) (2024)
- [HSIGene: A Foundation Model For Hyperspectral Image Generation](https://mlsanity.com/paper/2409.12470/) (2024)
- [Idempotent Generative Network](https://mlsanity.com/paper/2311.01462/) (2023)
- [A survey on GANs for computer vision: Recent research, analysis and taxonomy](https://mlsanity.com/paper/2203.11242/) (2022)
- [Metrics that matter: Evaluating image quality metrics for medical image generation](https://mlsanity.com/paper/2505.07175/) (2025)
- [StainDiffuser: MultiTask Dual Diffusion Model for Virtual Staining](https://mlsanity.com/paper/2403.11340/) (2024)
- [Score-Based Generative Models for Molecule Generation](https://mlsanity.com/paper/2203.04698/) (2022)
- [xTED: Cross-Domain Adaptation via Diffusion-Based Trajectory Editing](https://mlsanity.com/paper/2409.08687/) (2024)
- [Fine-tuning Language Models with Generative Adversarial Reward Modelling](https://mlsanity.com/paper/2305.06176/) (2023)
- [AutoMedPrompt: A New Framework for Optimizing LLM Medical Prompts Using Textual Gradients](https://mlsanity.com/paper/2502.15944/) (2025)
- [Collaborative Synthesis of Patient Records through Multi-Visit Health State Inference](https://mlsanity.com/paper/2312.14646/) (2023)
- [Generative Adversarial Networks for Image Augmentation in Agriculture: A Systematic Review](https://mlsanity.com/paper/2204.04707/) (2022)
- [Joint Semi-supervised 3D Super-Resolution and Segmentation with Mixed Adversarial Gaussian Domain Adaptation](https://mlsanity.com/paper/2107.07975/) (2021)
- [Score-based Idempotent Distillation of Diffusion Models](https://mlsanity.com/paper/2509.21470/) (2025)
- [SynQuE: Estimating Synthetic Dataset Quality Without Annotations](https://mlsanity.com/paper/2511.03928/) (2025)
- [Generative Adversarial Networks for Malware Detection: a Survey](https://mlsanity.com/paper/2302.08558/) (2023)
- [Regularized Generative Adversarial Network](https://mlsanity.com/paper/2102.04593/) (2021)
- [Learning Representative Trajectories of Dynamical Systems via Domain-Adaptive Imitation](https://mlsanity.com/paper/2304.10260/) (2023)
- [RDM: Recurrent Diffusion Model for Human Motion Generation](https://mlsanity.com/paper/2406.07169/) (2024)
- [LapLoss: Laplacian Pyramid-based Multiscale loss for Image Translation](https://mlsanity.com/paper/2503.05974/) (2025)
- [Beyond Anonymization: Object Scrubbing for Privacy-Preserving 2D and 3D Vision Tasks](https://mlsanity.com/paper/2504.16557/) (2025)
- [Generative diffusion model surrogates for mechanistic agent-based biological models](https://mlsanity.com/paper/2505.09630/) (2025)
- [IMPLICITSTAINER: Resolution Agnostic Data-Efficient Virtual Staining Using Neural Implicit Functions](https://mlsanity.com/paper/2505.09831/) (2025)
- [StolenLoRA: Exploring LoRA Extraction Attacks via Synthetic Data](https://mlsanity.com/paper/2509.23594/) (2025)
- [Diffusion-Based Synthetic Brightfield Microscopy Images for Enhanced Single Cell Detection](https://mlsanity.com/paper/2512.00078/) (2025)
- [A Wasserstein GAN-based climate scenario generator for risk management and insurance: the case of soil subsidence](https://mlsanity.com/paper/2605.06678/) (2026)
- [ROMS-IMLE: A Minimalist Approach to Competitive Single-Step Generative Modelling](https://mlsanity.com/paper/2607.19332/) (2026)

## Abstract (arXiv)

Generative adversarial networks (GANs) provide state-of-the-art results in image generation. However, despite being so powerful, they still remain very challenging to train. This is in particular caused by their highly non-convex optimization space leading to a number of instabilities. Among them, mode collapse stands out as one of the most daunting ones. This undesirable event occurs when the model can only fit a few modes of the data distribution, while ignoring the majority of them. In this work, we combat mode collapse using second-order gradient information. To do so, we analyse the loss surface through its Hessian eigenvalues, and show that mode collapse is related to the convergence towards sharp minima. In particular, we observe how the eigenvalues of the G are directly correlated with the occurrence of mode collapse. Finally, motivated by these findings, we design a new optimization algorithm called nudged-Adam (NuGAN) that uses spectral information to overcome mode collapse, leading to empirically more stable convergence properties.