I just read a few pages and find this thesis rather useful. Marked 🤓

In the last two years, a large set of papers enhanced GNNs' expressive power by marking subgraphs. In our recent NeurIPS 22 paper arxiv.org/abs/2206.11168, we link such approaches' expressive power to the k-WL and explore data-driven sampling techniques to select subgraphs.

I have added dozens of papers in my to-read list during the Learning on Graphs Conference 2024

You can sample nodes for scalable GNN training. But how do you do scalable inference? In our latest paper (Oral Learning on Graphs Conference 2024) we introduce influence-based mini-batching (IBMB) for both fast inference and training, achieving up to 130x and 17x speedups, respectively! 1/8 in 🧵

This is life 🤕

We'll present our work on data-driven graph rewiring at the #ICML23 Diff. Almost Everything workshop😄. If you're interested in how we tackle issues such as over-squashing and under-reaching, check our poster during session 2 at 4PM or virtually next week!

Recently, GNNs have been used to enhance exact solvers for hard optimization problems. In our new AISTATS paper (arxiv.org/abs/2310.10603), we show that GNNs can indeed solve linear optimization problems by simulating IPMs.

Ever wanted to rewire a graph but you couldn't decide on what heuristic to use👀? Fear not! Our #ICLR24 PR-MPNN method eliminates the need to ask ChatGPT for a list of interesting graph properties and learns how to automatically do rewiring in an end-to-end manner 🤯. 🧵 1/n

Imagine introducing our work at #AISTATS2024 when Prof Stephan Wright, whose book we base our theory on, comes to our poster 🤗 Bye bye Valencia ⛱️

I actually find it hard to contact the person on the other end of the table. Long range dependency…

Brainstorming at EEML with leading Google DeepMind research scientists on GNNs🕸️and neural algorithmic reasoning in Novi Sad 🇷🇸. w/ Petar Veličković Andreea Deac Chendi Qian Andrei Manolache Alicja Ziarko JJ Wilson Hải Yến Vũ Thị, Ognjen Stefanovic Stjepan Pozgaj Tigran Fahradyan @haririAli9