William Shatner, on his Blue Origin spaceflight: "It was among the strongest feelings of grief I have ever encountered. The contrast between the vicious coldness of space and the warm nurturing of Earth below filled me with overwhelming sadness." variety.com/2022/tv/news/w…

Elon Musk Jason DeBolt ⚡️ "Or, you know, when writing an article about me – I’ve had a long career in multiple industries. If you list my sins, I sound like the worst person on Earth. But if you put those against the things I’ve done right, it makes much more sense." - Elon Musk

I am excited to announce that our paper titled "sliced optimal partial transport arxiv.org/abs/2212.08049" has been accepted at #CVPR2025 #CVPR2023. Thanks to my mentor Dr. Soheil Kolouri, and collaborators Dr. Bernard Schmitzer, Dr. Mathew Thorpe. I am fortunate to work with all of you.

I am excited to share our #NeurIPS2023 paper "Energy-Based Sliced Wasserstein Distance". We introduce Energy-Based Sliced Wasserstein Distance (EBSW) which is an easy-to-implement and optimization-free variant of sliced Wasserstein with slicing distribution selection. (1/13)

I'm happy to share that our paper 'Quasi-Monte Carlo for 3D Sliced Wasserstein' (openreview.net/pdf?id=Wd47f7H…), where we empirically discuss the usage of spherical low-discrepancy point sets to approximate Sliced Wasserstein, has been accepted as a spotlight at #ICLR2024 (1/11).

I am excited to share our #ICLR2024 paper, "Sliced Wasserstein Estimation with Control Variates". We first introduce a variance reduction estimation of sliced Wasserstein distance. (1/11)

🏆🥇Happy to share that Stormer won the Best Paper Award at the Climate Change AI Workshop Climate Change AI #ICLR2024. Congrats to all my collaborators!

Finding incredible mentors is such a huge unlock

I’m feeling super happy and grateful to be able to share my work with all the amazing people at #CVPR2024. Seattle was such an incredible city. Nothing could have cheered me up more than the nice ocean breeze there 🌊

Happy to share our new ICML Conference #ICML2024 paper: "Sliced Wasserstein with Random-Path Projecting Directions", which introduces a new optimization-free projecting direction selection variant of sliced Wasserstein and application to training implicit diffusion models. (1/9)

Potential fields are a classical motion planning method, but suffers from local minima. At #ICML2024, we present a diffusion based potential field that helps avoid this issue! energy-based-model.github.io/potential-moti… Learned potentials can also be added together to solve harder test problems!

Happy Friday everyone ;)

I'm glad to share our #ICML2024 paper "Is Temperature Sample Efficient for Softmax Gaussian Mixture of Experts?". In this work, we study the effects of the dense-to-sparse (Gumble softmax) gate on the convergence of MoE and propose a novel gate to improve the performance. [1/n]

Introducing Streaming Diffusion Policy (SDP)! SDP allows you to generate actions using a single denoising step, with no distillation! SDP keeps an action buffer of partially denoised future actions, which is refined over time. streaming-diffusion-policy.github.io Led by Sigmund Hennum Høeg

Stormer is accepted at NeurIPS Conference 2024 🥳 Congrats to all authors!

Yet another super nice paper by Hyvarinen, how to implement Langevin when only having access to the score of a noised version of the target arxiv.org/abs/2410.05837

Good opportunity to announce Wasserstein Flow Matching (arxiv.org/abs/2411.00698)! New work in collaboration with Brandon Amos & Aram-Alexandre Pooladian on extending Flow Matching to the space of distributions!

Glad to announce that our work "Mirror and Preconditioned Gradient Descent in Wasserstein Space" was accepted at #NeurIPS2024 as a spotlight! This is a joint work with the amazing Théo Uscidda, A. David, P.C. Aubin-Frankowski and Anna Korba! Link: arxiv.org/abs/2406.08938

Excited to share our work on history guidance! By composing multiple diffusion score functions on different noise-corrupted versions of the history, we can construct a video generation model that can construct very long video rollouts and generalize well to OOD settings!

I am excited to share my new paper "Streaming Sliced Optimal Transport". I introduce the first algorithm for computing sliced Wasserstein distance from sample streams to make sliced optimal transport even faster and more scalable. (1/10)