Vignesh Kothapalli(@kvignesh1420) 's Twitter Profile Photo

Can 'Neural Collapse' occur in Graph Neural Networks?

Only if the computation graph satisfies a particular structural condition!

Checkout our paper (w/ Tom Tirer joan bruna) for additional NC insights on GNN training+inference on Stochastic Block Model graphs 🙂!

Can 'Neural Collapse' occur in Graph Neural Networks?

Only if the computation graph satisfies a particular structural condition! 

Checkout our #NeurIPS2023 paper (w/ @TirerTom @joanbruna) for additional NC insights on GNN training+inference on Stochastic Block Model graphs 🙂!
account_circle
Alfredo Canziani(@alfcnz) 's Twitter Profile Photo

Got to meet this one visualisation wizard in person! Truly impressive presentation.
My student got star struck as well! 😁
Thanks, Gabriel Peyré, for the amazing talk and being incredibly humble while listening to the feedback. Thanks, joan bruna, for inviting him!

Got to meet this one visualisation wizard in person! Truly impressive presentation.
My student got star struck as well! 😁
Thanks, @gabrielpeyre, for the amazing talk and being incredibly humble while listening to the feedback. Thanks, @joanbruna, for inviting him!
account_circle
Petar Veličković(@PetarV_93) 's Twitter Profile Photo

After 3 years, it's time for us to start sharing the chapters of the GDL book! ❤️

Also included: companion slides from our Cambridge University & University of Oxford courses 🧑‍🎓

Chapter 1 is out **now**!
More to follow soon 🎉

geometricdeeplearning.com/book 📖

Michael Bronstein joan bruna Taco Cohen

After 3 years, it's time for us to start sharing the chapters of the GDL book! ❤️

Also included: companion slides from our @Cambridge_Uni & @UniofOxford courses 🧑‍🎓

Chapter 1 is out **now**! 
More to follow soon 🎉

geometricdeeplearning.com/book 📖

@mmbronstein @joanbruna @TacoCohen
account_circle
ieva(@HyperboIeva) 's Twitter Profile Photo

Finally getting down to reading this fantastic thing from Michael Bronstein Petar Veličković Taco Cohen and joan bruna and I am struck by how, despite immediately jumping into specific mathematical detail, they define every new concept quickly and very clearly- I can always follow!

Finally getting down to reading this fantastic thing from @mmbronstein @PetarV_93 @TacoCohen and @joanbruna and I am struck by how, despite immediately jumping into specific mathematical detail, they define every new concept quickly and very clearly- I can always follow!
account_circle
Thomas Kipf(@tkipf) 's Twitter Profile Photo

Petar Veličković Cambridge University University of Oxford Michael Bronstein joan bruna Taco Cohen Congrats on reaching this milestone! Beautifully written, this will be a tremendously useful resource for the community and future students.

One nitpick: I wouldn’t necessarily bundle GCNs with spectral methods. It’s true that the GCN paper has an introductory section that

@PetarV_93 @Cambridge_Uni @UniofOxford @mmbronstein @joanbruna @TacoCohen Congrats on reaching this milestone! Beautifully written, this will be a tremendously useful resource for the community and future students.

One nitpick: I wouldn’t necessarily bundle GCNs with spectral methods. It’s true that the GCN paper has an introductory section that
account_circle
Petar Veličković(@PetarV_93) 's Twitter Profile Photo

Annual celebratory post: I retain my Affiliated Lectureship Cambridge University for 23/24! 🎉

Our GNN course will be significantly revamped, focusing on a Geometric Deep Learning perspective 🌐 following my book w/ Michael Bronstein joan bruna Taco Cohen

Details:
cl.cam.ac.uk/teaching/2324/…

Annual celebratory post: I retain my Affiliated Lectureship @Cambridge_Uni for 23/24! 🎉

Our GNN course will be significantly revamped, focusing on a Geometric Deep Learning perspective 🌐 following my book w/ @mmbronstein @joanbruna @TacoCohen

Details:
cl.cam.ac.uk/teaching/2324/…
account_circle
M²LInES(@M2LInES) 's Twitter Profile Photo

Our September newsletter is here! Featuring articles from Chris Pedersen, Pavel Perezhogin, joan bruna Laure Zanna, Griffin Mooers, and Marika Holland mailchi.mp/77c1eded6eb6/m…

Our September newsletter is here! Featuring articles from Chris Pedersen, Pavel Perezhogin, @joanbruna @laurezanna, Griffin Mooers, and Marika Holland   mailchi.mp/77c1eded6eb6/m…
account_circle
Petar Veličković(@PetarV_93) 's Twitter Profile Photo

For starters, why do we need a new framework?

Perhaps the most popular attempt towards a theory of architecture is Geometric DL (w/ Michael Bronstein joan bruna Taco Cohen). It leverages a perspective based on symmetry groups---but not all transformations of interest are symmetries!

For starters, why do we need a new framework?

Perhaps the most popular attempt towards a theory of architecture is Geometric DL (w/ @mmbronstein @joanbruna @TacoCohen). It leverages a perspective based on symmetry groups---but not all transformations of interest are symmetries!
account_circle
Alfredo Canziani(@alfcnz) 's Twitter Profile Photo

Congrats Min Jae Song for successfully passing your PhD Defense on «Continuous learning with errors and its applications», supervised by joan bruna and Oded Regev @ NYU Courant.

Congrats @mj_theory for successfully passing your PhD Defense on «Continuous learning with errors and its applications», supervised by @joanbruna and @regevlab @ @NYU_Courant.
account_circle
M²LInES(@M2LInES) 's Twitter Profile Photo

joan bruna is a co-author in this preprint, introducing Stochastic Optimal Control Matching (SOCM), a novel Iterative Diffusion Optimization (IDO) technique for stochastic optimal control. doi.org/10.48550/arXiv…

@joanbruna is a co-author in this preprint, introducing Stochastic Optimal Control Matching (SOCM), a novel Iterative Diffusion Optimization (IDO) technique for stochastic optimal control. doi.org/10.48550/arXiv…
account_circle