Chukwunonso presents at IEEE MECOM

A photo from the talk ft. Dr. Chukwunonso

This week Dr. Chukwunonso presented a cost-utility-based adversarial training approach for strengthening online AutoML classifiers against black-box evasation attacks in IoT networks. Presented at the IEEE Middle East Conference on Communications and Networking (MECOM 2026), the study demonstrated that Leveraging Bagging and Streaming Random Patches achieved 98.5% aversarial accuracy while maintaining 99.3% clean accuracy.