Latent Regularization in Generative Test Input Generation

TL;DR

This paper explores how regularizing latent spaces via truncation in style-based GANs enhances the quality of generated test inputs for deep learning classifiers, improving validity, diversity, and fault detection across multiple datasets.

Contribution

It introduces a latent code-mixing truncation strategy that outperforms random truncation in fault detection and input quality for deep learning classifier testing.

Findings

Latent code-mixing truncation improves fault detection rate.

The approach enhances diversity and validity of generated inputs.

Experimental results on MNIST, Fashion MNIST, and CIFAR-10 support these improvements.

Abstract

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.