Publications

Preprints and submitted papers

1. “Equivariant score-based generative models provably learn distributions with symmetries efficiently,” (with Ziyu Chen and Markos Katsoulakis), arXiv:2410.01244.

2. “Combining Wasserstein-1 and Wasserstein-2 proximals: robust manifold learning via well-posed generative flows,” (with Hyemin Gu, Markos Katsoulakis, and Luc Rey-Bellet), arXiv:2407:11901.

3. “Nonlinear denoising score matching for enhanced learning of structured distributions,” (with Jeremiah Birrell, Markos Katsoulakis, Luc Rey-Bellet, and Wei Zhu), arXiv:2405.15625.

4. “Wasserstein proximal operators describe score-based generative models and resolve memorization,” (with Siting Liu, Wuchen Li, Markos Katsoulakis, and Stanley Osher), arXiv:2402.06162.

5. “A mean-field games laboratory for generative modeling,” (with Markos Katsoulakis), arXiv:2304.13534.

6. “Computing the eigenfunctions of the multidimensional Ornstein-Uhlenbeck operator,” (with Tuhin Sahai and Youssef Marzouk) arXiv:2110.09229.

Journal articles

1. “Transport map unadjusted Langevin algorithms: learning and discretizing perturbed samplers,” (with Youssef Marzouk and Konstantinos Spiliopoulos), arXiv:2302.07227. Foundations of Data Science, accepted.

2. “Geometry-informed irreversible perturbations for accelerated convergence of Langevin dynamics,” (with Youssef Marzouk and Konstantinos Spiliopoulos), Statistics and Computing, 32(5), 78, 2022. Link arXiv:2108.08247.

3. “A Koopman framework for rare event simulation in stochastic differential equations,” (with Tuhin Sahai and Youssef Marzouk), Journal of Computational Physics 456 (2022): 111025. Link, arXiv:2101.07330.

4. “Suppression of epileptic seizures via Anderson localization,” (with Maysamreza Chamanzar and Mohammad-Reza Alam), Journal of The Royal Society Interface 14.127 (2017): 20160872. Link.

Peer-reviewed conference proceedings

1. “Score-based generative models are provably robust: an uncertainty quantification perspective,” (with Nikiforos Mimikos-Stamatopoulos and Markos Katsoulakis), arXiv:2405.15754. NeurIPS 2024, accepted.

Conference proceedings

1. “Sampling via Controlled Stochastic Dynamical Systems,” (with Tuhin Sahai and Youssef Marzouk), I (Still) Can’t Believe It’s Not Better! NeurIPS 2021 Workshop. 2021. Link.

2. “Rare event simulation of a rotorcraft system,” (with Youssef Marzouk, Byung-Young Min, and Tuhin Sahai), 2018 AIAA Non-Deterministic Approaches Conference. 2018. Link.

Theses

• “Efficient Sampling Methods of, by, and for Stochastic Dynamical Systems.” PhD Thesis, Massachusetts Institute of Technology, 2022. Link

• “A coupling approach to rare event simulation via dynamic importance sampling.” SM Thesis, Massachusetts Institute of Technology, 2017. Link

My Google Scholar page