Robust Bayesian Learning for Reliable Wireless AI: Framework and Applications

This work takes a critical look at the application of conventional machine learning methods to wireless communication problems through lens reliability and robustness. Deep techniques adopt frequentist framework, are known provide poorly calibrated decisions that do not reproduce true uncertainty caused by limitations in size training data. Bayesian learning, while principle capable addressing this shortcoming, is practice impaired model misspecification presence outliers. Both pervasive settings, which capacity models subject resource constraints data affected noise interference. In context, we explore framework robust learning. After tutorial-style introduction showcase merits on several important terms accuracy, calibration, robustness outliers misspecification.