Matthew E. Taylor

About

Matthew E. Taylor is a prominent researcher whose work spans reinforcement learning, transfer learning, and human-robot interaction. He has made foundational contributions to the field of transfer learning in reinforcement learning, most notably demonstrating how knowledge gained in one domain can accelerate learning in vastly different target tasks — work that has garnered over 200 citations and helped establish cross-domain transfer as a legitimate research frontier. His early investigations into inter-task mappings for policy search (133 citations) and rigorous comparisons of evolutionary versus temporal difference methods (88 citations) helped lay empirical groundwork for the broader reinforcement learning community. Taylor's research has increasingly focused on making AI systems collaborative with humans. His influential survey on Human-in-the-Loop reinforcement learning (131 citations) reframes RL as fundamentally a human-centered paradigm, while his work on integrating human demonstrations of varying ability (128 citations) and policy-dependent human feedback (108 citations) has advanced practical, interactive learning systems. His multi-task policy gradient research (149 citations) addresses critical sample-efficiency challenges in robotics. Bridging theory and application, Taylor has also explored physical robot transfer learning and smart home assistance technologies, reflecting a sustained commitment to deploying intelligent agents in real-world, human-serving environments.

Research Focus

Key Achievements

Top Papers

1. 1
   Cross-domain transfer for reinforcement learning

   204 citations · 2007
2. 2
   Online Multi-Task Learning for Policy Gradient Methods

   149 citations · 2014
3. 3
   Transfer via inter-task mappings in policy search reinforcement learning

   133 citations · 2007
4. 4
   Human-in-the-Loop Reinforcement Learning: A Survey and Position on Requirements, Challenges, and Opportunities

   131 citations · 2024
5. 5
   Integrating reinforcement learning with human demonstrations of varying ability

   128 citations · 2011
6. 6
   Robot-enabled support of daily activities in smart home environments

   110 citations · 2018
7. 7
   Interactive Learning from Policy-Dependent Human Feedback

   108 citations · 2017
8. 8
   Comparing evolutionary and temporal difference methods in a reinforcement learning domain

   88 citations · 2006
9. 9
   Transfer learning for reinforcement learning on a physical robot

   67 citations · 2010
10. 10
    DCOPs meet the realworld: exploring unknown reward matrices with applications to mobile sensor networks

    58 citations · 2009

Key Collaborators

Contact & Links