Hello world. I am generally interested in vision and time-series methods which invoke geometric or statistical principles to capture the physical structure in data (e.g., implicitly modeling directional distributions to model directional data). This is my North star. I favor these approaches over explicitly encoding physical constraints (e.g., energy conservation). Lately, these interests have manifested in work on diffusion models for scientific imagery domains. More to come.
I also enjoy trips to NYC for food and art destinations, which you’ll also occasionally see.
$$p(\rvx_t)$$