Towards Robust Graph Contrastive Learning

  • 2021-02-25 18:55:15
  • Nikola Jovanović, Zhao Meng, Lukas Faber, Roger Wattenhofer
  • 1

Abstract

We study the problem of adversarially robust self-supervised learning ongraphs. In the contrastive learning framework, we introduce a new method thatincreases the adversarial robustness of the learned representations through i)adversarial transformations and ii) transformations that not only remove butalso insert edges. We evaluate the learned representations in a preliminary setof experiments, obtaining promising results. We believe this work takes animportant step towards incorporating robustness as a viable auxiliary task ingraph contrastive learning.

 

Quick Read (beta)

loading the full paper ...