# PacGAN: The power of two samples in generative adversarial networks

arXiv: https://arxiv.org/abs/1712.04086 · HTML: https://mlsanity.com/paper/1712.04086/

## Key facts

- Cited by 104 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 12 Dec 2017.

## Built on by (most cited first, 30 of 104)

- [Flow Matching for Generative Modeling](https://mlsanity.com/paper/2210.02747/) (2022)
- [Tackling the Generative Learning Trilemma with Denoising Diffusion GANs](https://mlsanity.com/paper/2112.07804/) (2021)
- [MaxViT: Multi-Axis Vision Transformer](https://mlsanity.com/paper/2204.01697/) (2022)
- [CTAB-GAN: Effective Table Data Synthesizing](https://mlsanity.com/paper/2102.08369/) (2021)
- [Instance-Conditioned GAN](https://mlsanity.com/paper/2109.05070/) (2021)
- [CTAB-GAN+: Enhancing Tabular Data Synthesis](https://mlsanity.com/paper/2204.00401/) (2022)
- [Improved Transformer for High-Resolution GANs](https://mlsanity.com/paper/2106.07631/) (2021)
- [Generative Multiplane Images: Making a 2D GAN 3D-Aware](https://mlsanity.com/paper/2207.10642/) (2022)
- [TrajGAIL: Generating Urban Vehicle Trajectories using Generative Adversarial Imitation Learning](https://mlsanity.com/paper/2007.14189/) (2020)
- [Semi-Parametric Neural Image Synthesis](https://mlsanity.com/paper/2204.11824/) (2022)
- [RelaxLoss: Defending Membership Inference Attacks without Losing Utility](https://mlsanity.com/paper/2207.05801/) (2022)
- [How Realistic Is Your Synthetic Data? Constraining Deep Generative Models for Tabular Data](https://mlsanity.com/paper/2402.04823/) (2024)
- [A Comprehensive Survey of Synthetic Tabular Data Generation](https://mlsanity.com/paper/2504.16506/) (2025)
- [Multiresolution Equivariant Graph Variational Autoencoder](https://mlsanity.com/paper/2106.00967/) (2021)
- [Conditional GANs with Auxiliary Discriminative Classifier](https://mlsanity.com/paper/2107.10060/) (2021)
- [Attention Beats Concatenation for Conditioning Neural Fields](https://mlsanity.com/paper/2209.10684/) (2022)
- [How Do Flow Matching Models Memorize and Generalize in Sample Data Subspaces?](https://mlsanity.com/paper/2410.23594/) (2024)
- [Beyond the convexity assumption: Realistic tabular data generation under quantifier-free real linear constraints](https://mlsanity.com/paper/2502.18237/) (2025)
- [Closed-Loop Data Transcription to an LDR via Minimaxing Rate Reduction](https://mlsanity.com/paper/2111.06636/) (2021)
- [On the Privacy Properties of GAN-generated Samples](https://mlsanity.com/paper/2206.01349/) (2022)
- [Label-Efficient Deep Learning in Medical Image Analysis: Challenges and Future Directions](https://mlsanity.com/paper/2303.12484/) (2023)
- [Self-Diagnosing GAN: Diagnosing Underrepresented Samples in Generative Adversarial Networks](https://mlsanity.com/paper/2102.12033/) (2021)
- [MCL-GAN: Generative Adversarial Networks with Multiple Specialized Discriminators](https://mlsanity.com/paper/2107.07260/) (2021)
- [DPImageBench: A Unified Benchmark for Differentially Private Image Synthesis](https://mlsanity.com/paper/2503.14681/) (2025)
- [Diffusion-based generation of Histopathological Whole Slide Images at a Gigapixel scale](https://mlsanity.com/paper/2311.08199/) (2023)
- [TabularBench: Benchmarking Adversarial Robustness for Tabular Deep Learning in Real-world Use-cases](https://mlsanity.com/paper/2408.07579/) (2024)
- [Partition-Guided GANs](https://mlsanity.com/paper/2104.00816/) (2021)
- [Self-Supervised GANs with Label Augmentation](https://mlsanity.com/paper/2106.08601/) (2021)
- [The Representation Jensen-Shannon Divergence](https://mlsanity.com/paper/2305.16446/) (2023)
- [Disentanglement and Generalization Under Correlation Shifts](https://mlsanity.com/paper/2112.14754/) (2021)

## Abstract (arXiv)

Generative adversarial networks (GANs) are innovative techniques for learning generative models of complex data distributions from samples. Despite remarkable recent improvements in generating realistic images, one of their major shortcomings is the fact that in practice, they tend to produce samples with little diversity, even when trained on diverse datasets. This phenomenon, known as mode collapse, has been the main focus of several recent advances in GANs. Yet there is little understanding of why mode collapse happens and why existing approaches are able to mitigate mode collapse. We propose a principled approach to handling mode collapse, which we call packing. The main idea is to modify the discriminator to make decisions based on multiple samples from the same class, either real or artificially generated. We borrow analysis tools from binary hypothesis testing---in particular the seminal result of Blackwell [Bla53]---to prove a fundamental connection between packing and mode collapse. We show that packing naturally penalizes generators with mode collapse, thereby favoring generator distributions with less mode collapse during the training process. Numerical experiments on benchmark datasets suggests that packing provides significant improvements in practice as well.