Experiences Fall 2022 projects

Two of our ELLIS Excellence Fellows have published their great work at major AI conferences. Yannick Hogewind and Wietze Koops both did outstanding work in the area of learning and planning under uncertainty, in collaboration with Radboud AI, members of our ELLIS unit, and the Institute for Computing and Information Sciences.

• Safe Reinforcement Learning From Pixels Using a Stochastic Latent Representation. Yannick Hogewind, Thiago D. Simão, Tal Kachman, and Nils Jansen. In ICLR 2023.This paper proposes Safe SLAC, an algorithm that uses a stochastic latent variable model combined with a safety critic to address the problem of safe reinforcement learning in realistic, high-dimensional settings.
• Recursive Small-Step Multi-Agent A* for Dec-POMDPs. Wietze Koops, Nils Jansen, Sebastian Junges, and Thiago D. Simão. In IJCAI 2023. This paper presents a novel method for an exact algorithm to solve Dec-POMDPs, a typical model to capture multi-agent systems under uncertainty and partial observability. Their method shows superior performance on a rich set of standard benchmarks.