Diffusion models on Riemannian manifolds
I am extending traditional diffusion models on Euclidean spaces to Riemannian manifolds. I am interested in the efficient training and NN architectures of such models.
I am extending traditional diffusion models on Euclidean spaces to Riemannian manifolds. I am interested in the efficient training and NN architectures of such models.
Combining manifold-aware diffusion models with vision-language-action models in a hybrid approach to robot control.
2025
Preprint
Independent research paper
Using the framework of linear algebra and modular arithmetic to tackle an interesting problem of Knight Movement on the Chess Board.
2026
Preprint
Independent research paper
A paper asking how much of a stock's price movement is explained simply by the imbalance between buy and sell orders. Across 24 million quotes, the answer is about half. There maybe some non-linear dynamics better captured by modern ensemble methods.