Mixture of Experts for Node Classification

TL;DR

This paper introduces MoE-NP, a mixture of experts framework that enhances node classification by dynamically selecting specialized predictors based on node patterns, leading to improved accuracy on real-world datasets.

Contribution

The paper proposes a novel mixture of experts approach for node classification that adaptively combines predictors tailored to different node patterns, addressing limitations of uniform models.

Findings

MoE-NP outperforms existing methods on multiple datasets.

Dynamic model selection improves classification accuracy.

Significant performance gains demonstrate effectiveness.

Abstract

Nodes in the real-world graphs exhibit diverse patterns in numerous aspects, such as degree and homophily. However, most existent node predictors fail to capture a wide range of node patterns or to make predictions based on distinct node patterns, resulting in unsatisfactory classification performance. In this paper, we reveal that different node predictors are good at handling nodes with specific patterns and only apply one node predictor uniformly could lead to suboptimal result. To mitigate this gap, we propose a mixture of experts framework, MoE-NP, for node classification. Specifically, MoE-NP combines a mixture of node predictors and strategically selects models based on node patterns. Experimental results from a range of real-world datasets demonstrate significant performance improvements from MoE-NP.