Neural network models have become the leading solution for a large variety of tasks, such as classification, natural language processing, and others. However, their reliability is heavily plagued by adversarial inputs: inputs generated by adding tiny perturbations to correctly-classified inputs, and for which the neural network produces erroneous results. In this paper, we present a new method called Robustness Measurement and Assessment (RoMA), which measures the robustness of a neural network model against such adversarial inputs. Specifically, RoMA determines the probability that a random input perturbation might cause misclassification. The method allows us to provide formal guarantees regarding the expected frequency of errors that a trained model will encounter after deployment. The type of robustness assessment afforded by RoMA is inspired by state-of-the-art certification practices, and could constitute an important step toward integrating neural networks in safety-critical systems.
|Original language||American English|
|Title of host publication||Neural Information Processing - 29th International Conference, ICONIP 2022, Proceedings|
|Editors||Mohammad Tanveer, Sonali Agarwal, Seiichi Ozawa, Asif Ekbal, Adam Jatowt|
|Publisher||Springer Science and Business Media Deutschland GmbH|
|Number of pages||14|
|State||Published - 2023|
|Event||29th International Conference on Neural Information Processing, ICONIP 2022 - Virtual, Online|
Duration: 22 Nov 2022 → 26 Nov 2022
|Name||Communications in Computer and Information Science|
|Conference||29th International Conference on Neural Information Processing, ICONIP 2022|
|Period||22/11/22 → 26/11/22|
Bibliographical notePublisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
- Adversarial examples
- Neural networks