Latent Regularization in Generative Test Input Generation
This work addresses the challenge of generating high-quality test inputs for deep learning classifiers, which is crucial for reliability in domains like image classification, but it appears incremental as it builds on existing GAN methods with specific regularization tweaks.
This study tackled the problem of improving test input generation for deep learning classifiers by investigating latent regularization through truncation in style-based GANs, finding that a latent code-mixing approach increased fault detection rates and improved diversity and validity compared to random truncation.
This study investigates the impact of regularization of latent spaces through truncation on the quality of generated test inputs for deep learning classifiers. We evaluate this effect using style-based GANs, a state-of-the-art generative approach, and assess quality along three dimensions: validity, diversity, and fault detection. We evaluate our approach on the boundary testing of deep learning image classifiers across three datasets, MNIST, Fashion MNIST, and CIFAR-10. We compare two truncation strategies: latent code mixing with binary search optimization and random latent truncation for generative exploration. Our experiments show that the latent code-mixing approach yields a higher fault detection rate than random truncation, while also improving both diversity and validity.