Willis (Nanye) Ma

nm3607 [at] nyu [dot] edu

profile.jpg

I’m a senior student in Honors Mathematics and Computer Science at NYU Courant, where I’m fortunate to be advised by Saining Xie and work with Mark Goldstein, Michael Albergo, Nick Boffi, and Eric Vanden-Eijnden.

I’m broadly interested in Computer Vision, with a particular focus on measure transport in generative modeling. Here are some directions that I’m recently exploring:

Cross-domain Generation: solve measure transport between arbitrary densities to extend the scope of vision generative models beyond noise-to-image and to other modalities (speech-to-image, text-to-speech, etc.).

Resolving Real-world Problem with Limited Data: understand and leverage the knowledge learned by pre-trained vision generative models to approach various vision tasks in few- or zero-shot manners.

Long Form Video Generation: enhance the fidelity and consistency of video generation models, and further extend the spatiotemporal scale beyond the seconds scope.

Besides academics, I enjoy my leisure time in outdoor activities: basketball, hiking, and snowboarding (very new to this!), and I love using my camera to record moments of life. I’m also on my way to collecting playbooks of all Broadway shows (got six now!), and my favorite so far is Come from Away.

Here is my CV.

news

Jan 14, 2024 Release a blog post of our SiT!
Jan 14, 2024 Release the official PyTorch code of SiT! Check it out here SiT codebase
Dec 5, 2023 Glad to take part in Flatiron Institute’s workshop on Measure Transport, Sampling, and Diffusions!
Dec 4, 2023 Very grateful to be invited to participate in the NYC Vision Mini-workshop at Cornell Tech and learn from so many amazing projects!
Nov 18, 2023 Submitted our paper SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformers! :sparkles: Check it out here SiT preprint
Oct 13, 2023 Give a presentation on Diffusion and Flow models at Courant Summer Undergraduate Research Seminar! Here is the slides.