- Видео 7
- Просмотров 62 064
Marcus Brubaker
Добавлен 21 сен 2010
Assistant Professor of Computer Science at York University (Toronto) and Faculty Affiliate (Vector Institute).
Research interests include computer vision, machine learning and statistics.
Research interests include computer vision, machine learning and statistics.
Normalizing Flows and Invertible Neural Networks in Computer Vision (CVPR 2021 Tutorial)
CVPR 2021 Tutorial on Normalizing Flows and Invertible Neural Networks in Computer Vision
Looking for more about normalizing flows? Maybe start with these review articles: arxiv.org/abs/1908.09257 arxiv.org/abs/1912.02762
More information on the tutorial can be found here: mbrubake.github.io/cvpr2021-nf_in_cv-tutorial/
Looking for more about normalizing flows? Maybe start with these review articles: arxiv.org/abs/1908.09257 arxiv.org/abs/1912.02762
More information on the tutorial can be found here: mbrubake.github.io/cvpr2021-nf_in_cv-tutorial/
Просмотров: 16 669
Видео
Introduction to Normalizing Flows (ECCV2020 Tutorial)
Просмотров 39 тыс.4 года назад
A newer and more complete recording of this tutorial was made at CVPR 2021 and is available here: ruclips.net/video/8XufsgG066A/видео.html Tutorial on Normalizing Flows. Originally part of the ECCV2020 Tutorial on Normalizing Flows and Invertible Neural Networks in Computer Vision. Looking for more about normalizing flows? Maybe start with these review articles: arxiv.org/abs/1908.09257 arxiv.o...
Lost! Leveraging the Crowd for Probabilistic Visual Self-Localization
Просмотров 4,5 тыс.11 лет назад
Video of results for Brubaker et al "Lost! Leveraging the Crowd for Probabilistic Visual Self-Localization" in CVPR 2013. Abstract: In this paper we propose an affordable solution to self-localization, which utilizes visual odometry and road maps as the only inputs. To this end, we present a probabilistic model as well as an efficient approximate inference algorithm, which is able to utilize di...
Estimating Contact Dynamics (Results with Tracking)
Просмотров 30212 лет назад
The video of image-based tracking results corresponding to the research paper "Estimating Contact Dynamics." by Marcus A Brubaker, Leonid Sigal and David J Fleet. In Proceedings of IEEE ICCV 2009. Find out more at www.cs.toronto.edu/~mbrubake/
The Kneed Walker for Human Pose Tracking
Просмотров 36512 лет назад
The video corresponding to the research paper "The Kneed Walker for Human Pose Tracking." by Marcus A Brubaker and David J Fleet. In Proceedings of IEEE CVPR 2008. Find out more at www.cs.toronto.edu/~mbrubake/
Estimating Contact Dynamics (Results with Mocap)
Просмотров 37512 лет назад
The video of motion capture based results corresponding to the research paper "Estimating Contact Dynamics." by Marcus A Brubaker, Leonid Sigal and David J Fleet. In Proceedings of IEEE ICCV 2009. Find out more at www.cs.toronto.edu/~mbrubake/
Physics-based Person Tracking Using Simplified Lower Body Dynamics
Просмотров 52712 лет назад
The video corresponding to the research paper "Physics-based Person Tracking Using Simplified Lower Body Dynamics." by Marcus A Brubaker, David J Fleet and Aaron Hertzmann. In Proceedings of IEEE CVPR 2007. Find out more at www.cs.toronto.edu/~mbrubake/
00:00:00 - 1:12:30 - Introduction to Normalizing Flows 1:30:00 - Noise Flow 2:09:30 - Reacp about Normalizing Flows
3:02:13 for my reference cINN
At 47:00, the goal is to reduce dimensionality, but 4x4x1==2x2x4. How does that help?
The relevance of this tutorial from 2020 increased to the roof in 2024 after the recent release of Stable Diffusion 3 and its implementation of Flow Matching as an alternative to Diffusion. This is a very good building block to understand Flow Matching, that is why I happened here. It must be weird for researchers in Normalizing Flows at that time to witness the explosion of Gen AI through Diffusion models that were so close to what they were doing, it was like being missed by a nuclear bomb. But good research resist in the face of time and and the author predictions on Continuous-time Normalizing Flows and the research started by FFJORD were spot on. Kudos and thanks for putting this together, back in the day. I hope you resume posting videos like this!
First, explain the name, what is "normalizing"?
I will get out from this lecture that "If d is small that is not a big deal".
Thank you for making this excellent tutorial.
Around 19:30 you say that flow layers technically need to be diffeomorphisms, but is that actually the case? I don't see any reason why the inverse needs to be differentiable as well.
You need the flow to be differentiable in the normalizing direction in order to enable training and the computation of the Jacobian. Further, the Jacobian needs to be non-singular (non-zero determinant), and that implies (by the inverse function theorem) that the inverse is also differentiable.
@@MarcusBrubaker Ah, okay, makes sense. Thanks!
he keeps saying probabilistic graphical models when he meant probabilistic generative models
Thanks for the great explanatoin!
In coupling flows: does the split step split the x along the channel(features) or along others?
It can split the dimensions in any way. Traditionally in image applications the split is along channels dimensions, but, e.g., this paper split along spatial dimensions: proceedings.neurips.cc/paper/2020/hash/ecb9fe2fbb99c31f567e9823e884dbec-Abstract.html
this is a tutorial? xD
Thanks for the video! Very clearly explained :)
don't stop making videos, you are fantastic.
Shout-out to the dude brushing his teeth at 1:09:00 and contemplating whether he should leave his camera on or off
Super useful, I have been looking for this video for a long time, finally, I got it. Great video, keep going!
Glad it helped! You may want to check out the more recent version of this tutorial here: ruclips.net/video/8XufsgG066A/видео.html Much of the same content although a few things have been updated and refined.
Why GLOW is skipped??
Amazing video, will share it with my colleagues and friends, it deserves much more views than it has
Really like it :)
Thanks!
Great contents really appreciate it
thanks for the content
I have a question. How does taking cube root change a bimodal distribution to a unimodal distribution at 13:50?
It's hard to give a good intuitive explanation for how/why a cubic transform creates multi-modality. However. I can confirm that this is actually what happens in that particular example, those figures are the real result of transforming those distributions.
Thanks
Very nice slides and presentation.
WOW! This has been so helpful. You deserve a medal prof :)
Loved your tutorial, Marcus.
Great presentation, thanks for sharing it.
this video deserves a million view. so clearly explained.
Can we get the slides, please?
Nice tutorial.
awesome work
Really interesting idea! Thumbs up!