Brahma S. Pavse

Ph.D. Candidate, University of Wisconsin–Madison

pavse@wisc.edu Department of Computer Sciences, University of Wisconsin–Madison

I will be graduating in mid-2027 and will be on the job market for full-time research scientist roles. Please get in touch if you think I would be a good fit for your team!

I am a Computer Science Ph.D. candidate at the University of Wisconsin–Madison with a focus in machine learning and reinforcement learning, which is concerned with building agents that learn from experience. I am advised by Josiah Hanna, and have also worked closely with Yudong Chen and Qiaomin Xie. Previously, my research was supported by the Cisco Systems Distinguished Graduate Fellowship, and I was a research intern at Netflix Research where I worked with Qitong Gao, and at Sony AI where I worked with Varun Kompella.

Within reinforcement learning, I am interested in the broad area of representation learning, and what it means for an agent to learn good representations of its world for sequential decision-making. My work has shown that task-relevant representations can lead to higher quality evaluations of untested policies and stabilize value function learning [1, 2, 3] and that pre-trained vision foundation models can be leveraged to evaluate visuomotor robot control policies [4]. Currently, I am investigating how agents can perform hierarchical planning and hierarchical meta-reasoning with appropriate representations.

Previously, I completed my BS and MS in Computer Science from the University of Texas at Austin, where I was fortunate to be advised by Peter Stone. I also worked as a software engineer at Salesforce and SAS Institute.

Feel free to shoot me an email if you want to chat!

Selected Publications

RoboFQE: Rollout-Free Evaluation of Visual Robot Policies from Offline Trajectories
Yunfu Deng, Sangwoo Shin, Brahma S. Pavse, Josiah P Hanna
Conference on Robot Learning (CoRL), 2026

PDF Cite

@inproceedings{deng2026robofqe,
 author = {Deng, Yunfu and Shin, Sangwoo and Pavse, Brahma and Hanna, Josiah P},
 booktitle = {Conference on Robot Learning (CoRL)},
 organization = {PMLR},
 title = {RoboFQE: Rollout-Free Evaluation of Visual Robot Policies from Offline Trajectories},
 url = {https://arxiv.org/abs/2610.18699},
 year = {2026}
}

Stable Offline Value Function Learning with Bisimulation-based Representations
Brahma S. Pavse, Yudong Chen, Qiaomin Xie, Josiah P. Hanna
Proceedings of the 42nd International Conference on Machine Learning (ICML), 2025

PDF Cite

@inproceedings{pavse2025stable,
 archiveprefix = {arXiv},
 author = {Brahma S. Pavse and Yudong Chen and Qiaomin Xie and Josiah P. Hanna},
 booktitle = {Proceedings of the 42nd International Conference on Machine Learning (ICML)},
 eprint = {2410.01643},
 month = {July},
 primaryclass = {cs.LG},
 title = {Stable Offline Value Function Learning with Bisimulation-based Representations},
 url = {https://arxiv.org/abs/2410.01643},
 year = {2025}
}

State-Action Similarity-Based Representations for Off-Policy Evaluation
Brahma S. Pavse, Josiah P. Hanna
Advances in Neural Information Processing Systems (NeurIPS), 2023

PDF Code Cite

@inproceedings{pavse2023rope,
 archiveprefix = {arXiv},
 author = {Brahma S. Pavse and Josiah P. Hanna},
 booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
 eprint = {2310.18409},
 month = {December},
 primaryclass = {cs.LG},
 title = {State-Action Similarity-Based Representations for Off-Policy Evaluation},
 url = {https://arxiv.org/abs/2310.18409},
 year = {2023}
}

Scaling Marginalized Importance Sampling to High-Dimensional State-Spaces via State Abstraction
Brahma S. Pavse, Josiah P. Hanna
Proceedings of the 37th AAAI Conference on Artificial Intelligence (AAAI), 2023 (Oral presentation)

PDF Cite

@inproceedings{pavse2023absmis,
 address = {Washington, DC, USA},
 author = {Brahma S. Pavse and Josiah P. Hanna},
 booktitle = {Proceedings of the 37th AAAI Conference on Artificial Intelligence (AAAI)},
 month = {February},
 title = {Scaling Marginalized Importance Sampling to High-Dimensional State-Spaces via State Abstraction},
 url = {documents/papers/2023/pavse_2023_absmis.pdf},
 year = {2023}
}

Reinforced Inverse Dynamics Modeling for Learning from a Single Observed Demonstration
Brahma S. Pavse, Faraz Torabi, Josiah Hanna, Garrett Warnell, Peter Stone
IEEE Robotics and Automation Letters (RA-L); IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2020

PDF DOI Cite

@article{pavse2020ridm,
 author = {Brahma S. Pavse and Faraz Torabi and Josiah Hanna and Garrett Warnell and Peter Stone},
 doi = {10.1109/LRA.2020.3010750},
 journal = {IEEE Robotics and Automation Letters (RA-L); IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
 month = {July},
 title = {Reinforced Inverse Dynamics Modeling for Learning from a Single Observed Demonstration},
 url = {documents/papers/2020/pavse_2020_ridm.pdf},
 year = {2020}
}

Reducing Sampling Error in Batch Temporal Difference Learning
Brahma S. Pavse, Ishan Durugkar, Josiah Hanna, Peter Stone
Proceedings of the 37th International Conference on Machine Learning (ICML), 2020

PDF Cite

@inproceedings{pavse2020psec,
 author = {Brahma S. Pavse and Ishan Durugkar and Josiah Hanna and Peter Stone},
 booktitle = {Proceedings of the 37th International Conference on Machine Learning (ICML)},
 month = {July},
 title = {Reducing Sampling Error in Batch Temporal Difference Learning},
 url = {documents/papers/2020/pavse_2020_psec.pdf},
 year = {2020}
}

Brahma S. Pavse

News

Selected Publications