Recent Publication at VLDB ’26

A big shot out to Dr. Devarakonda! His paper, entitled “Scalable GNN Explanations with Distributed Shapley Values” has been accepted to the 52nd International Conference on Very Large Data Bases (VLDB) 2026!

The paper, “Scalable GNN Explanations with Distributed Shapley Values”, is a collaboration with Ariful Azad (Texas A&M) and Selahattin Akkas (Meta Platforms). The paper tackles the problem of scaling distributed-memory Shapley values computation for GNN explainability. This work leverages a distributed CGLS solver to scale computation to 128 GPUs and is the first to tackle GNN problems with millions of features.

Congratulations!