GANetic Loss for Generative Adversarial Networks with a Focus on Medical Applications

TL;DR

This work tackles the instability and mode-collapse challenges in GAN training by automatically designing loss functions with Genetic Programming. The authors evolve a loss family on a simple DCGAN using MNIST, identifying GANetic as a robust candidate, defined by a data-driven expression that emphasizes correct real-class classification and regularization-like behavior. GANetic is then evaluated across CIFAR10 and multiple medical-imaging tasks, improving image generation quality (FID/KID) and anomaly-detection performance (AUROC) while delivering stable, reproducible training. The results demonstrate a practical, data-driven path toward more reliable GAN training with clear applicability to medical image generation and biomarker discovery.

Abstract

Generative adversarial networks (GANs) are machine learning models that are used to estimate the underlying statistical structure of a given dataset and as a result can be used for a variety of tasks such as image generation or anomaly detection. Despite their initial simplicity, designing an effective loss function for training GANs remains challenging, and various loss functions have been proposed aiming to improve the performance and stability of the generative models. In this study, loss function design for GANs is presented as an optimization problem solved using the genetic programming (GP) approach. Initial experiments were carried out using small Deep Convolutional GAN (DCGAN) model and the MNIST dataset, in order to search experimentally for an improved loss function. The functions found were evaluated on CIFAR10, with the best function, named GANetic loss, showing exceptionally better performance and stability compared to the losses commonly used for GAN training. To further evalute its general applicability on more challenging problems, GANetic loss was applied for two medical applications: image generation and anomaly detection. Experiments were performed with histopathological, gastrointestinal or glaucoma images to evaluate the GANetic loss in medical image generation, resulting in improved image quality compared to the baseline models. The GANetic Loss used for polyp and glaucoma images showed a strong improvement in the detection of anomalies. In summary, the GANetic loss function was evaluated on multiple datasets and applications where it consistently outperforms alternative loss functions. Moreover, GANetic loss leads to stable training and reproducible results, a known weak spot of GANs.

Figures (8)

• Figure 1: Implemented GP search process. In the first row an example of the generated trees is shown, in this example $x$ and $y$ are $y_{real}$ and $y_{pred}$.
• Figure 2: Variation of results on CIFAR10 for different losses over 25 runs.
• Figure 3: Examples of images generated by prGANs using different losses for BreCaHAD (left), Hyper-Kvasir (middle) and LAG (right) datasets. From top to bottom: ground truth images, original loss and GANetic loss.
• Figure 4: GANetic and BCE loss functions (left) and their gradients (right).
• Figure E: Examples of images generated by DCGANs for CIFAR10 dataset. GANs trained by BCE (top) or trained by GANetic loss (bottom).