I am interested in developing machine intelligence. My research sits at the intersection of reinforcement learning and statistics, and focuses on how we can use statistical tools to improve generalization in reinforcement learning. My most recent paper, Invariance Through Inference, provides a good overview of this setting. For more thoughts on generalization, see also my recent papers on planning in robotics and causal inference.
I am a research assistant professor at TTIC, an academic institute located on the campus of the University of Chicago. In 2018, I completed my PhD at UC Berkeley. My advisor was Pieter Abbeel. In 2016 and 2017, I was a research scientist at Open AI, where I was advised by Ilya Sutskever. I received a BA in mathematics from the University of Chicago, where I spent four wonderful years. During this time, I had the honor of working under Paul Sally.
I am on the academic job market this year.
My Google Scholar page can be found here.
My CV is here.
Invariance Through Inference
Takuma Yoneda, Ge Yang, Matthew Walter, Bradly C. Stadie
Submitted to International Conference in Learning Representations (ICLR) 2022. Preprint here
World Model as a Graph: Learning Latent Landmarks for Planning
Lunjun Zhang, Ge Yang, Bradly C. Stadie
In International Conference on Machine Learning (ICML), Long Presentation, July 2021. Site here
Maximum Entropy Gain Exploration for Long Horizon Multi-goal Reinforcement Learning
Silviu Pitis, Harris Chan, Stephen Zhao, Bradly Stadie, Jimmy Ba
In International Conference on Machine Learning, July 2020. Paper here
Learning Intrinsic Rewards as a Bi-Level Optimization Problem
Lunjun Zhang, Bradly C. Stadie, Jimmy Ba
Conference on Uncertainty in Artificial Intelligence (UAI), July 2020. Paper here
One-Shot Pruning of Recurrent Neural Networks by Jacobian Spectrum Evaluation
Matthew Zhang, Bradly C. Stadie
In International Conference on Learning Representations (ICLR), Addis Ababa, Ethiopia, April 2020. ArXiv link
Transfer Learning for Estimating Causal Effects Using Neural Networks
Bradly C. Stadie, Soeren R. Kuenzel, Nikita Vemuri, Varsha Ramakrishnan, Jasjeet S. Sekhon, Pieter Abbeel
INFORMS Annual Meeting ML and causal inference workshop (2019). ArXiv link
One Demonstration Imitation Learning Bradly C. Stadie, Siyan Zhao, Qiqi Xu, Bonnie Li, Lunjun Zhang
Preprint. See here
Evolved Policy Gradients
Rein Houthooft, Richard Y. Chen, Phillip Isola, Bradly C. Stadie, Filip Wolski, Jonathan Ho, Pieter Abbeel
In Neural Information Processing Systems (NeurIPS) [Spotlight], Montreal, Canada, December 2018. ArXiv
Some Considerations on Learning to Explore via Meta-Reinforcement Learning
Bradly C. Stadie, Ge Yang, Rein Houthooft, Xi Chen, Yan Duan, Yuhuai Wu, Pieter Abbeel, Ilya Sutskever
In Neural Information Processing Systems (NeurIPS), Montreal, Canada, December 2018. ArXiv
Learning to Learn from Flawed, Failed, and Figurative Demonstrations
Ge Yang, Bradly C. Stadie, Roberto Calandra, Pieter Abbeel, Sergey Levine, Chelsea Finn
In Neural Information Processing Systems (NeurIPS) Deep RL workshop [Spotlight], Montreal, Canada, December 2018. Paper here.
Third-Person Imitation Learning
Bradly C. Stadie, Pieter Abbeel, Ilya Sutskever
In the proceedings of the International Conference on Learning Representations (ICLR), Toulon, France, April 2017. ArXiv
One-Shot Imitation Learning
Yan (Rocky) Duan, Marcin Andrychowicz, Bradly C. Stadie, Jonathan Ho, Jonas Schneider, Ilya Sutskever, Pieter Abbeel, Wojciech Zaremba
In Neural Information Processing Systems (NeurIPS), Long Beach, California, December 2017. ArXiv
Incentivizing Exploration in Reinforcement Learning with Deep Predictive Models
Bradly C. Stadie, Sergey Levine, Pieter Abbeel
In Neural Information Processing Systems (NeurIPS) Deep RL Workshop, Montreal, Canada, December 2015 ArXiv