Dr. Yangchen Pan is a Departmental Lecturer in Machine Learning at the University of Oxford's Department of Engineering Science. His research focuses on achieving sample-efficient generalization in machine learning, particularly in settings involving distribution shifts (e.g., adversarial learning, domain adaptation) and adaptive capabilities like offline/online reinforcement learning and continual learning. He has contributed to foundational work in reinforcement learning, robustness, and risk-averse optimization. Education and Academic Background: While specific degree details are not explicitly listed, his academic trajectory includes roles at leading institutions such as the University of Alberta (PhD, 2017-2020), where he collaborated with prominent researchers like Martha White and Amir-massoud Farahmand. Additional affiliations include teaching roles at the University of Waterloo, University of Toronto, and Indiana University. Research Interests: Pan's work spans machine learning theory and applications, with emphasis on scalable algorithms for complex decision-making systems. Key areas include adversarial robustness, offline reinforcement learning, and mitigating distribution shifts in real-world deployments. His recent contributions explore risk-aware policy optimization and novel approaches to sample efficiency. Professional Service: Pan serves on the program committees of top conferences like NeurIPS, ICML, and ICLR. He has reviewed for journals including the Journal of Machine Learning Research and Transactions on Machine Learning Research. Teaching: Pan teaches advanced courses in optimization, machine learning, and AI at the University of Oxford, including C25 Optimization and AI/ML with Python. He has also taught at the University of Alberta and Indiana University. Labs/Teams: While no specific lab name is mentioned, his research is closely tied to Oxford's Engineering Science department and collaborative projects with institutions like the ZERO Institute (as seen in event leadership at IMAD2025).
