Risk-Averse Certification of Bayesian Neural Networks
This work addresses robustness certification for Bayesian neural networks, which is important for safety-critical applications, but it appears incremental as it builds on existing certification methods by adding risk measures.
The paper tackles the problem of evaluating robustness in Bayesian neural networks by proposing a Risk-Averse Certification framework (RAC-BNN) that integrates Conditional Value at Risk (CVaR) to provide probabilistic guarantees, resulting in tighter certified bounds and higher efficiency in complex tasks.
In light of the inherently complex and dynamic nature of real-world environments, incorporating risk measures is crucial for the robustness evaluation of deep learning models. In this work, we propose a Risk-Averse Certification framework for Bayesian neural networks called RAC-BNN. Our method leverages sampling and optimisation to compute a sound approximation of the output set of a BNN, represented using a set of template polytopes. To enhance robustness evaluation, we integrate a coherent distortion risk measure--Conditional Value at Risk (CVaR)--into the certification framework, providing probabilistic guarantees based on empirical distributions obtained through sampling. We validate RAC-BNN on a range of regression and classification benchmarks and compare its performance with a state-of-the-art method. The results show that RAC-BNN effectively quantifies robustness under worst-performing risky scenarios, and achieves tighter certified bounds and higher efficiency in complex tasks.