Introducing Sequential Predictive Conformal Inference (SPCI)

🔍 Talk Abstract: Are you interested in cutting-edge developments in time series analysis? Join us to explore Sequential Predictive Conformal Inference (SPCI), a groundbreaking distribution-free conformal prediction algorithm for sequential data, such as time series. SPCI takes into account the non-exchangeable nature of time series data, making it highly relevant in situations where existing conformal prediction methods fall short. Chen Xu will delve into the intricacies of this innovative approach, which adaptively re-estimates conditional quantiles of non-conformity scores by leveraging temporal dependencies. Learn how SPCI can significantly reduce prediction interval widths compared to other methods while maintaining the desired empirical coverage. This is a talk you won't want to miss! 🎙️ About the Speaker: Chen Xu, a 4th-year PhD student in Operations Research at the Georgia Institute of Technology, is at the forefront of research in optimization, machine learning, and statistics. His expertise includes distribution-free uncertainty quantification, particularly in conformal prediction, for complex models. Additionally, Chen is passionate about flow-based generative neural networks as scalable computational tools, addressing challenges in diverse applications ranging from renewable energy and wildfire monitoring to medical diagnosis and transportation.