Fabrizio Frasca

Community Takes on the Future of Graph Learning On drifting goals, fresh frontiers, and a bittersweet lesson we are slowly learning

Check out this post from @ffabffrasca.bsky.social and the GLOW reading group on the future of graph learning!

I’ve also contributed and my main take is - its actually working and its an exciting moment to work on applications!

open.substack.com/pub/glowread...

After a break in April we welcome all of you to the next session of GLOW🌟 next week!

Join and interact with our speakers Christian Koke (scale in GNNs) and Yonatan Sverdlov (sparse geometric MPNNs).

🗓️ May 28th, 5pm CEST on Zoom.

🌐 Details & sign-up: sites.google.com/view/graph-learning-on-weds.

In arxiv.org/abs/2412.07106, we leverage modern graph similarity—capturing the fine-grained geometry of MPNNs' feature space—to derive generalization bounds for MPNNs. Our theory is broad, covering various aggregation and loss functions.

My former KTH colleague Ioana Bercea is hiring for two Ph.D. students in (randomized) algorithms and data structures. This is a great opportunity embedded into a very nice TCS group at KTH!
www.kth.se/lediga-jobb/...

Want to know about the current understanding of the generalization abilities of GNNs? Please have a look at our survey paper arxiv.org/abs/2503.15650.

Joint work with Antonis Vasileiou, Stefanie Jegelka, and Ron Levie.

GLOW is returning on 𝗠𝗮𝗿𝗰𝗵 𝟮𝟲𝘁𝗵, 𝟱𝗽𝗺 𝗖𝗘𝗧 with a special guest: @petar-v.bsky.social 🌟

He will lecture on LLMs as GNNs – a topic which received quite some attention at our last session.

Specifically, we will learn how Graph ML tools can help understand LLM generalisation

New blog post on overcoming academic impostor syndrome. This arose from great discussions during our group retreat, and I decided to make it publicly available in the hope of providing some utility to #academia.

#AcademicSky #AcademicChatter

bastian.rieck.me/blog/2025/im...

The Spring issue of the Bulletin of the EATCS is now available at www.eatcs.org/images/bulle...

In the TCS on the Web Column, I had the pleasure to talk to Rachel Cummings and Sumegha Garg, looking behind the scenes of TCS+ @tcsplus.bsky.social.

Apply if you are working on Geometric Deep Learning!

Had such a great experience as a student in the past year.

This year I will be joining as a mentor along with so many other great researchers!

Hope to see you in London ☺️

I am excited to host the next edition of the NeurReps Global Seminar Series, from @neurreps.bsky.social!
@ninamiolane.bsky.social and @louisacornelis.bsky.social will be speaking on 🧠Geometric Intelligence in Minds and Machines🤖
📅Friday, Feb 21st, 11am PST
Join us at UCSB (Elings,1605) or on Zoom!

I am hiring for a Ph.D. student to work in the areas of social network analysis, algorithms and fair machine learning.

Please apply and join our highly motivated team.

For more information please see the call: neumannstefan.com/hiring/

Don’t miss out on this one :)

Make sure to tune in today at 17 CET 🌟

I am honestly so surprised about how happy people are with ChatGPT for coding and maths. In the last couple of days, I asked it to implement some quite basic algorithms and maths questions, and each time it did not do particularly well and most times I gave up after a couple of minutes.

Join us in 30 mins! 🌟

Join this year's last edition of Graph Learning on Wednesdays (GLOW) tomorrow at 5pm CET!

Visual depiction of r-lGIN: During preprocessing, we calculate the path neighborhoods Nr (v) for each node v in the graph G. Paths of varying lengths are processed separately using simple GINs, and their embeddings are pooled to obtain the final graph embedding. The forward complexity scales linearly with the sizes of Nr (v), enabling efficient computation on sparse graphs.

Today at NeurIPS:

Weisfeiler and Leman go Loopy: A New Hierarchy for Graph Representational Learning

Cycles are important for predictive tasks on chemical molecules. We allow message passing along neighboring paths. Our architecture can subgraph-count cycles and homomorphism-count cactus graphs.

LOG24 - Italy meetup - Organisers Pietro Bongini PostDoc @ Unisi

Such an honour to hold a keynote talk in Siena this morning at the #LoG Meetup!

Wonderful time in a beautiful city with a super community 🖤

Grateful to @cate-graz.bsky.social and the rest of the team (sites.google.com/student.unis...) for making this happen!

As last time, the talks will be 15 minutes each, with plenty of time for discussion in breakout rooms afterwards!

Join us via Zoom on Dec 18th @ 17 CET ...

... and stay tuned by subscribing to our mailing list: groups.google.com/g/graph-lear...

Understanding Virtual Nodes: Oversmoothing, Oversquashing, and Node Heterogeneity Message passing neural networks (MPNNs) have been shown to have limitations in terms of expressivity and modeling long-range interactions. Augmenting MPNNs with a virtual node (VN) removes the localit...

Afterwards, Josh will help us to better understand the benefits of virtual nodes in MPNNs to mitigate problems such as over-smoothing and over-squashing: arxiv.org/abs/2405.13526.

Homomorphism Counts for Graph Neural Networks: All About That Basis A large body of work has investigated the properties of graph neural networks and identified several limitations, particularly pertaining to their expressive power. Their inability to count certain pa...

First, we will hear from Emily about how to properly include homomorphism counts into graph features to make MPNNs more expressive: arxiv.org/abs/2402.08595.

🌟 GLOW is coming back in December with amazing speakers: Emily Jin and @joshsouthern.bsky.social !

🗓️ Dec 18th @ 17 CET on Zoom, don't miss that!

🌐 Find more here: sites.google.com/view/graph-l...

We’re closing out the @logconference.bsky.social meetup in Aachen with a panel on the relationship between graph learning and computer vision with fantastic panelists 🥳

Stephan Guennemann kicking off the LOG local meetup in Aachen.

Topological & geometrical machine learning starter pack
go.bsky.app/HnBdLJU

Screen capture of LoG website. The conference is Nov 26-29, 2024. See our website for more details: https://logconference.org/.

✨EXCITING NEWS! We now have a Bluesky account!!

AND registration for the 3rd Learning on Graphs conference is open 😊

It is virtual, free to attend, livestreamed, and recorded 📹 Sign up today! The conference is this coming week 😱

forms.gle/eYiDCopJGUc8...

Today’s session was super 🌟

Thanks so much to our speakers Maya and Paolo and to everyone who attended.

Looking fwd to our next meeting — stay tuned!