Graph Neural Networks

Exploring graph-based deep learning

I am interested in the applications of graph neural networks to domains like physics and social networks. I worked on graph neural networks for particle track reconstruction at CERN and on modeling information spread on social networks as part of SimPPL. I also used graph based deep learning to design multimodal recommender systems at Adobe Research.

Visual Depiction of the Graph Neural Network Architecture I developed at Adobe Research

  • I delivered a talk on Graph Neural Networks and another on Interpretable ML to the BFS Reading Group at the NYU Courant Institute; presenting both at Unicode Research as well

  • After remotely collaborating for nearly a year, I am spending May 2022 in the UK with Oxford’s Torr Vision Group visiting Prof. Philip Torr and Prof. Atilim Gunes Baydin to work on social networks, disinformation, and recommendation systems using graph neural networks and continual learning to expand this paper.