BIU learning club – Chaim Baskin – Graph Representation Learning Through Recoverability

Zoom link: 

https://us02web.zoom.us/j/4685913265

Title:

Graph Representation Learning Through Recoverability 

Abstract:

Self-supervised learning methods became popular for graph representation learning because they do not rely on manual labels and offer better generalization. Contrastive methods based on mutual information maximization between augmented instances of the same object are widely used in self-supervised learning of representations. For graph-structured data, however, there are two obstacles to successfully utilizing contrastive methods: the data augmentation strategy and training decoder for mutual information estimation between augmented representations of nodes, sub-graphs, or graphs. In this work, we propose a self-supervised graph representation learning algorithm, Graph Information Representation Learning (GIRL). GIRL does not require augmentations or a decoder for mutual information estimation. The algorithm is based on an alternative information metric, recoverability, which is tightly related to mutual information but is less complicated when estimating. Our self-supervised algorithm consistently outperforms existing state-of-the-art contrast-based self-supervised methods by a large margin on a variety of datasets. In addition, we show how the recoverability can be used in a supervised setting to alleviate the effect of over-smoothing/squashing in deeper graph neural networks.

Short Bio:

Dr. Chaim Baskin is an adjunct Faculty member/Senior Research Associate at the CS department and an academic member of Tech-AI (Technion Artificial Intelligence Hub) at Technion-IIT. Chaim holds a Ph.D. in Machine Learning from the CS department, Technion. Previously he was a post-doc fellow in the same department. His industry experience includes Deep Learning research positions at Intel and Intel Labs. He published more than Twenty papers and preprints in multiple fields of deep learning, including neural network compression, self-supervised learning, computer vision, and Graph Neural Networks. Chaim acted as PI/co-PI/Lead researcher at several successful research collaborations with leading companies such as Nvidia, Intel, Hyundai, Ceva, Mafat, Elta, etc.