Mathematics Colloquia and Seminars

Return to Colloquia & Seminar listing

Machine learning has drawn much recent attention in its ability to model, predict, and control network performance and user experience, based on network measurement and user performance data. In this talk, we introduce both applied and theoretical aspects of machine learning algorithms in networks. First, we discuss our recent work on learning-based cellular resource management, using the example of collaborative-learning-based cellular tower parameter configuration. Second, we present our proposed opportunistic learning algorithms for multi-armed bandits and episodic reinforcement learning that are inspired by real-world applications of learning algorithms in networks.