Marcus Brubaker

We also present a solution to the problem, showing how we can efficiently estimate 2nd order information and largely mitigate the poor conditioning, opening the door for efficient, SGD-based high resolution cryo-EM methods.

We show the numerical conditioning of cryo-EM reconstruction is increasingly poor at high resolution such that SGD has no real hope of resolving high resolution structures.

Ever wonder why SGD, which works so well for ab initio cryo-EM, never seems to get to high resolution on its own? Well, we have the answer: conditioning!

Calling all PhD students in vision/ML interested in working with a great team (+me!) at GDM doing cutting edge research in 3D computer vision and generative models!

Definitely. None of those things automatically go away just because someone gives you a degree!

Congratulations to Shayan for a great job presenting our work at the poster on Friday and in a talk @workshopmlsb.bsky.social today!

Come check out our work on semi-amortized inference for cryoEM today at NeurIPS!

📅 Friday, Dec 13, 11 AM - 2 PM
📌 East Exhibit Hall A-C, Poster #1105

Paper: arxiv.org/abs/2406.10455
Website: shekshaa.github.io/semi-amortized-cryoem/

With: Shayan Shekarforoush, David Lindell and David Fleet

A very similar argument (start with a basic set of requirements/axioms) can be made to show that probabilities are basically the only logically consistent calculus of uncertainty, see en.wikipedia.org/wiki/Cox%27s...

Heading to Vancouver for #NeurIPS this week. If you're around and interested in catching up, feel free to DM me!

Ever tried to run COLMAP or other SfM pipeline only to have it fail due to movement in the video? Check out our new work exploring on how detect and mask dynamic objects!

Anne Gagneux, Ségolène Martin, @quentinbertrand.bsky.social Remi Emonet and I wrote a tutorial blog post on flow matching: dl.heeere.com/conditional-... with lots of illustrations and intuition!

We got this idea after their cool work on improving Plug and Play with FM: arxiv.org/abs/2410.02423

😅

Writing a good scientific paper

For those who missed this post on the-network-that-is-not-to-be-named, I made public my "secrets" for writing a good CVPR paper (or any scientific paper). I've compiled these tips of many years. It's long but hopefully it helps people write better papers. perceiving-systems.blog/en/post/writ...

📢 Research Opportunities!

We’re seeking highly motivated postdocs & PhD students to dive into groundbreaking #ComputerVision research, exploring the future of 3D/4D neural rendering.

1/3

GitHub - gle-bellier/flow-matching: Annotated Flow Matching paper Annotated Flow Matching paper. Contribute to gle-bellier/flow-matching development by creating an account on GitHub.

I created 3 introductory notebooks on Flow Matching models to help get started with this exciting topic! ✨

1. Annotated Flow Matching paper: github.com/gle-bellier/...
2. Discrete Flow Matching: github.com/gle-bellier/...
3. Minimal FM in Jax: github.com/gle-bellier/...

Can you add me and @csprofkgd.bsky.social to AI?

Welcome!

Can Computers Create Art? This essay discusses whether computers, using Artificial Intelligence (AI), could create art. First, the history of technologies that automated aspects of art is surveyed, including photography and an...

6 years ago, in the days of GAN art, I wrote an article called "Can Computers Create Art?", arguing that computers should not be considered artists, regardless of how good image generation gets. People sometimes ask if my views have changed. I say, 1/🧵
www.mdpi.com/2076-0752/7/...

YouTube video by Aaron Hertzmann SIGGRAPH 2024 Award Talk: My journey in art and computing (so far)

Here's my the talk for the #SIGGRAPH 2024 Computer Graphics Achievement award, about my experiences in painting, computer graphics and the science of art. And how so much art and research are unplanned and unpredictable.
www.youtube.com/watch?v=ehgj...
aaronhertzmann.com/2024/08/19/j...

The Computer Vision Foundation open access proceedings team is proud to announce that the #CVPR2024 proceedings is now online:

Main conference: openaccess.thecvf.com/CVPR2024

Workshops: openaccess.thecvf.com/CVPR2024_wor...

Enjoy and I'll see all of you in Seattle!

Congrats! Looks very interesting, will add it to my reading list