Distribution Free Prediction Sets for Node Classification

TL;DR

This paper introduces a method to create reliable prediction sets for node classification in graph neural networks by adapting conformal prediction techniques to account for graph dependencies, resulting in more accurate uncertainty quantification.

Contribution

It develops a novel conformal prediction approach tailored for GNNs that considers graph structure, improving the calibration and tightness of prediction sets.

Findings

Provides tighter, better calibrated prediction sets than naive methods

Demonstrates effectiveness on benchmark GNN datasets

Adapts conformal prediction to dependent data in graphs

Abstract

Graph Neural Networks (GNNs) are able to achieve high classification accuracy on many important real world datasets, but provide no rigorous notion of predictive uncertainty. Quantifying the confidence of GNN models is difficult due to the dependence between datapoints induced by the graph structure. We leverage recent advances in conformal prediction to construct prediction sets for node classification in inductive learning scenarios. We do this by taking an existing approach for conformal classification that relies on \textit{exchangeable} data and modifying it by appropriately weighting the conformal scores to reflect the network structure. We show through experiments on standard benchmark datasets using popular GNN models that our approach provides tighter and better calibrated prediction sets than a naive application of conformal prediction.