Gergely Neu

3 papers Universitat Pompeu Fabra

• Distances for Markov chains from sample streams
• Offline imitation learning in $Q^\pi$-realizable MDPs without expert realizability
• Sparse Optimistic Information Directed Sampling