Montaser Mohammedalamen, PhD

"Generalization in Monitored Markov Decision Processes (Mon-MDPs)"
Conference Talk — Reinforcement Learning Conference (RLC)
Montreal, Canada  ·  2026

Investigates how Mon-MDP agents generalize across environments with varying reward observability patterns, establishing theoretical conditions and empirical benchmarks for robust policy transfer beyond training distributions.

watch paper code

"Learning to Be Cautious"
Paper Talk — Transactions on Machine Learning Research (TMLR)
2025

Presents the minimax regret framework for training RL agents that behave conservatively in novel, out-of-distribution states without requiring human supervision or prior knowledge of failure modes.

watch paper

"Monitored Markov Decision Processes"
Conference Talk — AAMAS 2024
Auckland, New Zealand  ·  May 2024

Introduces the Mon-MDP framework, a formal treatment of RL environments where reward signals are not always observable, and analyses the theoretical and practical consequences for agent behavior.

watch paper

"Breaking the Wall of Inaccessibility: Wheelchair Robot Controlled by Brain Signal"
Falling Walls Lab Finals
Berlin, Germany  ·  Nov 2017  ·  Audience Prize Winner

Pitch presentation at the international Falling Walls Lab Finals in Berlin, demonstrating a wheelchair robot controlled entirely by EEG brain signals — awarded the Audience Prize among finalists from around the world.

watch