Teaching

Assistant Professor @ Illinois Tech

CS 595: Decentralized Machine Learning Systems

This is an original course based around many topics related to my research in federated learning, decentralized learning, edge computing, Internet-of-Things, etc.

Description: The vast majority of data are naturally decentralized. Modern approaches to Artificial Intelligence (AI) — which are trained on large-scale, centralized infrastructure—are insufficient to learn from the majority of data available. To this end, machine learning methods that can span from the cloud to the network edge are needed. This course will investigate the state-of-the-art literature for learning across decentralized data. This course will largely focus on AI on edge and fog computing systems. The course will include lectures on subject areas that include (but are not limited to) edge/fog computing, complex systems, machine learning, federated learning, split learning, gossip learning, system and statistical heterogeneity, and knowledge aggregation methods. The course will also task students to design and complete their own project to investigate a research question related to the course material. Student-led paper discussions will also be had to learn from the latest state-of-the-art.