Koulik Khamaru

Koulik Khamaru is an Assistant Professor of Statistics at Rutgers University, located at Hill Center 403. His research focuses on the theory and application of statistics, machine learning, and optimization, with specific interests in EM algorithms, Gaussian mixture models, reinforcement learning, and non-convex optimization.

Education:

• PhD in Statistics from University of California, Berkeley (advisors: Martin J. Wainwright and Michael I. Jordan)
• Undergraduate and Master's degrees in Statistics from Indian Statistical Institute, Kolkata

Research Interests:

• Statistical theory of adaptive models and algorithms
• Reinforcement learning and sequential decision-making
• Optimization in high-dimensional and non-convex settings
• Algorithmic convergence and statistical guarantees

Recent Work: Recent papers include advancements in adaptive linear estimating equations (NeurIPS 2023), instance-dependent reinforcement learning analysis (JMLR), and variance-reduced stochastic approximation techniques. His work is supported by NSF grant DMS-2311304.

Teaching: Taught a course on Reinforcement Learning in Spring 2024.