Boosting Ai Security For Robust, Reliable Systems

Researchers at EPFL have developed a new training approach to ensure that machine learning models, particularly deep neural networks, consistently perform as intended, significantly enhancing their reliability. This approach replaces a long-standing approach to training based on zero-sum game, and employs a continuously adaptive attack strategy to create a more intelligent training scenario. The research was awarded an esteemed Best Paper Award at the 2023 International Conference on Machine Learning’s New Frontiers and Adversarial Machine Learning Workshop.