معرفی
Steve Hanneke is an Assistant Professor in the Computer Science Department at Purdue University, specializing in statistical learning theory, machine learning, and algorithmic information theory. His research focuses on understanding the fundamental limits of learning from data, including questions about what can be learned and how efficiently it can be done. Prior to Purdue, he held positions at Toyota Technological Institute at Chicago (2018–2021), Carnegie Mellon University (2009–2012), and Princeton University (2018 visiting lecturer). He earned his PhD from Carnegie Mellon University in 2009, advised by Eric Xing and Larry Wasserman, with a thesis on active learning foundations.
Key research interests include active learning, adversarial robustness, online learning, and the theoretical analysis of learning algorithms. He has contributed to foundational work on PAC learning, sample complexity, and universal learning frameworks. Notable awards include the Best Paper Awards at ALT 2021 and COLT 2020, and his 2007 ICML paper received an Honorable Mention for the ICML Test of Time Award in 2017.
Teaching experience includes courses at Purdue (Machine Learning Theory, Data Mining and Machine Learning), Princeton (Statistical Learning and Nonparametric Estimation), and Carnegie Mellon (Advanced Probability and Statistical Theory). His work has been published in top venues like COLT, NeurIPS, and the Journal of Machine Learning Research, with over 50 peer-reviewed articles.
Research highlights include developing the theory of universal learning under general stochastic processes, characterizing minimax rates in active and online learning, and exploring adversarial robustness in PAC learning frameworks. Current projects focus on bandit learning, non-stationary environments, and the theoretical limits of learning algorithms.
