Papers
36
Total Citations
8,419
H-Index
23
About
Marc Peter Deisenroth is a leading researcher at the intersection of machine learning, robotics, and reinforcement learning, whose work has fundamentally shaped how autonomous systems learn from limited data. Best known for his highly influential 2017 survey "Deep Reinforcement Learning: A Brief Survey," which has accumulated over 4,200 citations, Deisenroth has helped define and communicate the trajectory of modern AI to researchers worldwide. His contributions span both theoretical foundations and practical applications, with a particular emphasis on data-efficient learning — enabling robots to acquire skills without requiring vast amounts of costly real-world experience. A hallmark of his research is the application of Gaussian Processes to robotics and control, most notably through his PILCO framework, which demonstrated that probabilistic model-based reinforcement learning could drastically reduce the number of trials needed for a robot to master a task. His survey on policy search for robotics (684 citations) remains a foundational reference in the field. Deisenroth has also advanced multi-task learning and human-robot interaction, broadening the scope of adaptive autonomous systems. Collectively, his work bridges elegant probabilistic methodology with real-world robotic challenges, making him an indispensable figure in contemporary machine learning research.
Research Focus
Key Achievements
Top Papers
- 1Deep Reinforcement Learning: A Brief Survey4,261 citations · 2017
- 2A Brief Survey of Deep Reinforcement Learning754 citations · 2017
- 3A Survey on Policy Search for Robotics684 citations · 2011
- 4Gaussian Processes for Data-Efficient Learning in Robotics and Control656 citations · 2013
- 5Bayesian optimization for learning gaits under uncertainty290 citations · 2015
- 6
- 7Manifold Gaussian Processes for regression214 citations · 2016
- 8
- 9Multi-task policy search for robotics121 citations · 2014
- 10