Petar Veličković
@PetarV_93

Senior Research Scientist @DeepMind | Affiliated Lecturer @Cambridge_Uni. PhD @TrinCollCam. Ex-@Mila_Quebec @BellLabs. GDL, GNNs, Algorithms. 🇷🇸🇲🇪🇧🇦




Petar Veličković    @PetarV_93
Proud to announce that for the 2021-22 academic year, I will be teaching a full 16h Master's course on Graph Representation Learning @Cambridge_CL (with @pl219_Cambridge)! Accordingly, @Cambridge_Uni has granted me the title of Affiliated Lecturer, starting this October! 😳🥳

Petar Veličković    @PetarV_93
We're looking to expand our team for core graph representation learning @DeepMind, with particular emphasis on GNNs! Plentiful exciting opportunities for both research and applied work. If interested, check out the link in @PeterWBattaglia's tweet + feel free to reach out!

Petar Veličković    @PetarV_93
Started listening to @BartoszMilewski's Category Theory course on YouTube (https://t.co/gENxWhJkEl). 45 minutes in, I'm now left wondering why anybody ever wants to study anything other than Category Theory. 😶 Superb delivery in a manner very accessible to programmers!

Petar Veličković    @PetarV_93
Lastly, I joined @logml2021 as a project mentor (steering a mini-project which unifies graph neural networks, classical algorithms, and text adventures). I found the opportunity of tightly collaborating with attendees very attractive, and I look forward to the result! (5/9)

Petar Veličković    @PetarV_93
At @EEMLcommunity 2021, I will give a lecture on graph neural networks from the ground up, followed by a GNN lab session led by @ni_jovanovic. I will also host a mentorship session with several aspiring mentees! Based on 2020, I anticipate a recording will be available! (2/9)

Petar Veličković    @PetarV_93
At this year's PSI:ML seminar (powered by @MicrosoftSrbija et al.), I will deliver a longer introductory lecture on GNNs and geometric DL. PSI:ML is the prime machine learning summer school for undergrads in my home country, and I am honoured to be invited to speak there. (4/9)

Petar Veličković    @PetarV_93
I firmly believe in giving back to the community I came from, as well as paying forward and making (geometric) deep learning more inclusive to underrepresented communities in general. Accordingly, this summer you can (virtually) find me on several summer schools! A thread (1/9)

Petar Veličković    @PetarV_93
Our latest work (lead by Jonathan Godwin) on very deep GNNs, scaled up to 100 layers with monotonic improvements! This has been made possible with a novel choice of denoising, and block-sharing structure between the layers. We obtain strong results on both OpenCatalyst and QM9.

Petar Veličković    @PetarV_93
Happy to announce that EPMP has now been accepted at the #ICML2021 Workshop on Computational Biology! It feels great to come back to the venue that supported our (AG-)Fast-Parapred line of models three years ago, and we look forward to discussing our work with everyone!

Petar Veličković    @PetarV_93
When I first presented GATs and node classification to a general machine learning audience, I kept using analogies like 'DL on steroids', or, 'in the real world, your data-points are linked'. Recent paper from @OATML_Oxford now makes this link explicit! https://t.co/mJIBBMYsH6
 Reply      Retweet   31      Like    219   

Petar Veličković    @PetarV_93
Important read of the day: GATv2 (Brody, @urialon1, @yahave): https://t.co/p8SXyQF7Go The exact attention mechanism I used in the GAT paper was intentionally 'weakened' to make it work on the easy-to-overfit datasets of the time. It was never meant to be a 'silver bullet'... 1/2
 Reply      Retweet   22      Like    86   

Petar Veličković    @PetarV_93
Life after #NeurIPS











 








Petar Veličković

Facebook AI

DeepMind

Sergey Levine

Yann LeCun

Andrew Gordon Wilson