Tight Verification of Probabilistic Robustness in Bayesian Neural Networks

arXiv:2401.11627v2 Announce Type: replace
Abstract: We introduce two algorithms for computing tight guarantees on the probabilistic robustness of Bayesian Neural Networks (BNNs). Computing robustness guarantees for BNNs is a significantly more challenging task than verifying the robustness of standard Neural Networks (NNs) because it requires searching the parameters' space for safe weights. Moreover, tight and complete approaches for the verification of standard NNs, such as those based on Mixed-Integer Linear Programming (MILP), cannot be directly used for the verification of …

abstract algorithms arxiv bayesian computing cs.ai cs.fl cs.lg cs.lo networks neural networks nns parameters robustness searching space standard type verification