公开(公告)号：WO2022117760A1

公开(公告)日：2022-06-09

申请号：PCT/EP2021/084042

申请日：2021-12-02

申请人： IMPERIAL COLLEGE INNOVATIONS LIMITED

发明人： LOMUSCIO, Alessio ,  HENRIKSEN, Patrick ,  KOUVAROS, Panagiotis

IPC分类号： G06N3/04 ,  G06N3/08 ,  G06N5/00

摘要： Systems and methods are provided for verifying the transformational robustness of a neural network. Data is obtained representing a trained neural network, a set of algebraic constraints on the output of the network, and a range of inputs to the neural network over which the algebraic constraints are to be verified, such that the data defines a transformational robustness verification problem. A set of complementary constrains on the pre-activation of a node in the network are then determined such that for any input in the range of inputs, at least one of the complementary constraints is satisfied. A plurality of child verification problems are generated based on the transformational robustness verification problem and the set of complementary constraints. For each child verification problem, it is determined whether a counter-example to the child verification problem exists. Based on the determination of whether counter-examples to the child verification problems exist, it is determined whether the neural network is transformationally robust.