Meet our AI-powered robot that’s ready to play table tennis. 🤖🏓 It’s the first agent to achieve amateur human level performance in this sport. Here’s how it works. 🧵

Generally: a neural network is equivariant if, when its input changes in a predictable way, its output must change in a predictable way as well. Attached: a beautiful example from Yann LeCun's LeNet. As you shift the '4', all the NN's representations shift together with it.

The thrill of starting to write experiments for a new project is unmatched 🤌

I rarely post sports content here but I honestly shouted at the screen this evening... 🥲 great job guys. you almost had them. idemo dalje 🫡🏀

Still one of my personal favourite papers I co-authored, whose insights I use in my research to this day 🧑‍🔬 Thanks Learning on Graphs Conference 2024 for giving EGP an epic home ❤️ [++ EGP won the NeurIPS'22 GLFrontiers Best Paper Award, which is one of the academic rewards I cherish the most 😊]

The single most undervalued fact of linear algebra: Matrices are graphs, and graphs are matrices. Encoding matrices as graphs is a cheat code, making complex behavior simple to study. Let me show you how!

It's been a pleasure to teach at #EEML24 EEML this Summer! The students' and organisers' affection was almost unbearable! 🥹 Thanks, Petar Veličković, for thinking of me as opening speaker and equipment testing Guinea pig 🐹 I hope you enjoyed the opening stand-up comedy! 🎭

Meet Gemini Live: a new way to have more natural conversations with Gemini. 💬 💡 Brainstorm ideas ❓ Interrupt to ask questions ⏸️ Pause a chat and come back to it Now rolling out in English to Gemini Advanced subscribers on Android phones → dpmd.ai/46RToL9

This summer Novi Sad 🇷🇸 was the epicentre of the global AI community, hosting EEML 2024! 🌍 310 participants from 47 countries (840 applications!) 🧑‍🏫 Lectures on deep learning, VLMs, RL etc from top AI experts 💻 Tutorial sessions 🛜 Networking and research discussions

My paper "Training-free Graph Neural Networks and the Power of Labels as Features" has been accepted to #TMLR 🎉 I proposed training-free (and optionally trained) GNNs. Paper📜：arxiv.org/abs/2404.19288 Code📁：github.com/joisino/laf

Continuing the very happy tradition 🎉🧑‍🏫 I retain my affiliated lectureship ("adjunct professorship" for friends in US/Canada) Cambridge Computer Science for a 4th consecutive year! ❤️ Looking fwd to teaching Geometric DL 🌐 with Pietro Lio' to an amazing Master's student cohort again!

The paper that kickstarted the graph representation learning revolution! 🕸️Congrats Bryan Perozzi on the Test of Time award for DeepWalk -- super well deserved! 👏👏

Oldies but goldies: Volker Strassen, Gaussian Elimination is not Optimal, 1969. Strassen algorithm reduces the complexity of matrix multiplication in terms of number of multiplications. en.wikipedia.org/wiki/Strassen_…

Topcoder paused their competitions, so I had to go for another 4000.

About 20h left to submit your abstract for the greatest graph event of the year! 🚀

such moments are few and far between in research, but when they happen, they fill the soul with warmth. Preprint coming soon 👀

We’re presenting AlphaProteo: an AI system for designing novel proteins that bind more successfully to target molecules. 🧬 It could help scientists better understand how biological systems function, save time in research, advance drug design and more. 🧵 dpmd.ai/3XuMqbX