My research focuses on how AI can learn through interaction with humans. Virtually all applications of AI involve solving problems defined by humans. The goal of my work is to allow humans and artificial agents to work together to build a shared understanding of such tasks, enabling AI that leverages human knowledge, and ensuring that learned behaviors are truly aligned with human intentions. My research attempts to cover all aspects of this problem, from its theoretical foundations to the development of scalable interactive learning algorithms, and their evaluation with real human users. My work spans many topics within and related to AI, including (Deep) Reinforcement Learning, Game Theory and Multi-Agent Systems, and Cognitive Science.
I am currently a Lecturer in Machine Learning at the University of Sheffield. I received my PhD in Computer Science from North Carolina State University in 2019, under the supervision of Dave Roberts. My dissertation examined the types of latent knowledge conveyed by human-provided feedback and demonstrations, and developed interactive learning algorithms which leverage models of human behavior to extract this information. I received my Bachelor's in Computer Science from Georgia Tech in 2011. After completing my PhD, I did a two-year post-doc with Microsoft Research Cambridge, exploring the use of Reinforcement Learning and Interactive Learning in commercial game development. I also completed a post-doc at TU Delft with Dr. Frans Oliehoek, applying game theory to human-AI cooperation.
R. Loftin and F. A. Oliehoek. "On the Impossibility of Learning to Cooperate with Adaptive Partner Strategies in Repeated Games," in International Conference on Machine Learning, 2022, pp. 14197-14209. Bibtex
copy
@inproceedings{loftin2022impossibility,
title={On the Impossibility of Learning to Cooperate with Adaptive Partner Strategies in Repeated Games},
author={Loftin, Robert and Oliehoek, Frans A.},
booktitle={International Conference on Machine Learning},
pages={14197--14209},
year={2022}
}
R. Loftin, A. Saha, S. Devlin, K. Hofmann. "Strategically Efficient Exploration in Competitive Multi-agent Reinforcement Learning," in the 37th Conference on Uncertainty in Artificial Intelligence, 2021. Bibtex
copy
@inproceedings{loftin2021strategic,
title={Strategically Efficient Exploration in Competitive Multi-agent Reinforcement Learning},
author={Loftin, Robert and Saha, Aadirupa and Devlin, Sam and Hofmann, Katja},
booktitle={the 37th Conference on Uncertainty in Artificial Intelligence},
year={2021}
}
R. Loftin, B. Peng, J. MacGlashan, M. L. Littman, M. E. Taylor, J. Huang, D. L. Roberts. "Learning Behaviors via Human-Delivered Discrete Feedback: Modeling Implicit Feedback Strategies to Speed Up Learning," Autonomous Agents and Multi-Agent Systems, vol. 30, issue 1, pp. 30-59, Jan 2016. Bibtex
copy
@article{loftin2016sabl,
author={Loftin, Robert and Peng, Bei and MacGlashan, James and Littman, Michael L. and Taylor, Matthew E. and Huang, Jeff and Roberts, David L.},
title={Learning behaviors via human-delivered discrete feedback: modeling implicit feedback strategies to speed up learning},
journal={Autonomous Agents and Multi-Agent Systems},
year={2016},
month={Jan},
volume={30},
number={1},
pages={30--59},
issn={1573-7454},
doi={10.1007/s10458-015-9283-7}
}
R. Loftin, J. MacGlashan, B. Peng, M. E. Taylor, M. L. Littman, D. L. Roberts. "A Strategy-Aware Technique for Learning Behaviors from Discrete Human Feedback," in the Twenty-Ninth AAAI Conference on Artificial Intelligence, 2014, pp. 937-943. Bibtex
copy
@inproceedings{loftin2014aaai,
title={A Strategy-Aware Technique for Learning Behaviors from Discrete Human Feedback},
author={Loftin, Robert and MacGlashan, James and Peng, Bei and Taylor, Matthew E and Littman, Michael L and Huang, Jeff and Roberts, David L},
booktitle={the Twenty-Ninth AAAI Conference on Artificial Intelligence},
pages={937--943},
year={2014}
}
