Summary of Gnnshap: Scalable and Accurate Gnn Explanation Using Shapley Values, by Selahattin Akkas and Ariful Azad
GNNShap: Scalable and Accurate GNN Explanation using Shapley Values
by Selahattin Akkas, Ariful Azad
First submitted to arxiv on: 9 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Social and Information Networks (cs.SI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes GNNShap, a novel approach to explaining graph neural networks (GNNs) by utilizing game-theoretic Shapley value methods. Current explanation techniques for GNNs are considered black box models and lack interpretability. Existing Shapley value-based approaches have limitations in handling large-scale graphs due to computational costs. To address these challenges, the authors introduce a parallelized sampling strategy on GPUs and batching model predictions to accelerate explanations. The proposed method, GNNShap, provides better fidelity scores and faster explanations compared to baselines on real-world datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine trying to understand how a computer program works without knowing why it makes certain decisions. This is like a “black box”! Graphs are important in many scientific fields, but the models that process them (called GNNs) can be mysterious too. The researchers behind this paper want to make these models more understandable by using a technique called Shapley value. They developed a new way to use Shapley value, which they call GNNShap, to explain how GNNs work. This approach is faster and better than existing methods on real-world data. |
