Boya Hou
I am a postdoctoral research associate at the University of Illinois Urbana-Champaign (UIUC), working with Prof. Maxim Raginsky and Prof. Olgica Milenkovic. Prior to this, I received my Ph.D in Electrical and Computer Engineering at UIUC advised by Prof. Subhonmesh Bose. I obtained B.Eng. in Electrical Engineering from Zhejiang University.
Research
My research focuses on developing machine learning methods for complex systems across science and engineering, spanning learning, probabilistic sampling, control, and stability analysis. Drawing on machine learning, operator theory, control and dynamical systems, I take a unified approach to these problems through learned operator representations that govern the evolution of functions or distributions. These operators are linear even when the underlying system is nonlinear. I construct them from trajectory data with finite-sample guarantees, and exploit their structure to design efficient algorithms for each task, with provable guarantees that black-box models do not offer. Active research directions include:
- Operator learning with finite-sample guarantees: learning operator representations of nonlinear dynamical systems from trajectory data, with compression, multi-agent, and online algorithms that scale to large, distributed and streaming data, applied to data-driven decision-making.
- Operator-guided probabilistic sampling: sampling trajectories of diffusion processes by integrating operator learning with optimal stochastic control, with applications to transition path sampling in machine learning and molecular dynamics.
- Applications in scientific AI and engineering: molecular and cellular dynamics, analysis of drug intervention, and power grid operations.
- Dynamical systems for AI: e.g., data attribution and unlearning for LLMs.
Publications
* co-first authors
📄 Nonparametric Sparse Online Learning of the Koopman Operator.
To appear in SIAM Journal on Control and Optimization (SICON), 2026 download.
B. Hou, S. Sanjari, N. Dahlin, A. Koppel, S. Bose
📄 Sparse Learning of Dynamical Systems in Reproducing Kernel Hilbert Space: An Operator-Theoretic Approach.
Proceedings of the 40th International Conference on Machine Learning (ICML), 2023 download.
B. Hou, S. Sanjari, N. Dahlin, S. Bose, U. Vaidya
📄 Transition Path Sampling Using Koopman Operators and Exit-Time Optimal Control.
In submission, 2026. Short version accepted at a NeurIPS 2026 workshop. download.
B. Hou*, S. Wang*, S. Ambekar, M. Raginsky, O. Milenkovic
📄 Spatially-Coupled Network RNA Velocities: A Control-Theoretic Perspective.
In submission, 2026. download.
B. Hou, M. Raginsky, A. Pandey, O. Milenkovic
Journal Articles
- Nonparametric Sparse Online Learning of the Koopman Operator
To appear in SIAM Journal on Control and Optimization (SICON), 2026 · Preprint
B. Hou, S. Sanjari, N. Dahlin, A. Koppel, S. Bose - Kernel Methods for the Approximation of the Eigenfunctions of the Koopman Operator
Physica D: Nonlinear Phenomena, 2025 · Download
J. Lee, B. Hamzi, B. Hou, H. Owhadi, G. Santin, U. Vaidya - Data-Driven Transient Stability Analysis Using the Koopman Operator
International Journal of Electrical Power and Energy Systems, 2024 · Download
A. Reddy Ramapuram Matavalam, B. Hou, H. Choi, S. Bose, U. Vaidya - Propagating Uncertainty Through System Dynamics in Reproducing Kernel Hilbert Space
Physica D: Nonlinear Phenomena, 2024 · Download
B. Hou, A. Reddy Ramapuram Matavalam, S. Bose, U. Vaidya - Impact of Aviation Electrification on Airports: Flight Scheduling and Charging
IEEE Transactions on Intelligent Transportation Systems, 2023 · Download
B. Hou, S. Bose, L. Marla, K. Haran
Conference Proceedings
- Sparse Learning of Dynamical Systems in Reproducing Kernel Hilbert Space: An Operator-Theoretic Approach
Proceedings of the 40th International Conference on Machine Learning (ICML), 2023 · Download
B. Hou, S. Sanjari, N. Dahlin, S. Bose, U. Vaidya - Compressed Decentralized Learning of Conditional Mean Embedding Operators in Reproducing Kernel Hilbert Space
Proceedings of the 37th AAAI Conference on Artificial Intelligence, 2023 · Download
B. Hou, S. Sanjari, N. Dahlin, S. Bose - Sparse Learning of Kernel Transfer Operators
Asilomar Conference on Signals, Systems, and Computers, 2021 · Download
B. Hou, S. Bose, U. Vaidya - Powering Electric Aircraft at O’Hare Airport: A Case Study
IEEE PES General Meeting, 2020 · Download
B. Hou, S. Bose, K. Haran
Preprints
- Transition Path Sampling Using Koopman Operators and Exit-Time Optimal Control
In submission, 2026. Short version accepted at a NeurIPS 2026 workshop · Preprint
B. Hou*, S. Wang*, S. Ambekar, M. Raginsky, O. Milenkovic - Spatially-Coupled Network RNA Velocities: A Control-Theoretic Perspective
In submission, 2026 · Preprint
B. Hou, M. Raginsky, A. Pandey, O. Milenkovic - A Behavioral Framework for Data-Driven Modeling of Nonlinear Systems in Vector-Valued Reproducing Kernel Hilbert Spaces
Under revision, 2026 · Preprint
B. Hou, M. Raginsky - GUARD: Guided Unlearning and Retention via Data Attribution for Large Language Models
In submission, 2026
P. Niu*, E. Ma*, B. Hou*, H. Zhou, D. Zhou, H. Zhang, R. Etesami, O. Milenkovic
Selected Awards
- Best Poster Award, Power Systems Engineering Research Center (PSERC), 2024
- Rising Stars in EECS, 2023
- M.A.Pai Scholarship, 2023
- Mavis Future Faculty Fellows (MF3), 2023-2024
Fun
Outside of mathematics and engineering, I pursue my passion for dancing. I performed contemporary ballet and dance of the Han-Tang dynasties (a subcategory of Classical Chinese Dance) with the Champaign Park District and Zhejiang Wenqin Art Troupe.