Graph Sparsification for Enhanced Conformal Prediction in Graph Neural Networks
This work addresses the problem of enhancing reliability and efficiency in conformal prediction for graph neural networks, representing an incremental improvement over existing post-hoc methods.
The paper tackled the challenge of improving conformal prediction during the training stage for graph neural networks by introducing SparGCP, which uses graph sparsification and a conformal prediction-specific objective, resulting in an average reduction of prediction set sizes by 32%.
Conformal Prediction is a robust framework that ensures reliable coverage across machine learning tasks. Although recent studies have applied conformal prediction to graph neural networks, they have largely emphasized post-hoc prediction set generation. Improving conformal prediction during the training stage remains unaddressed. In this work, we tackle this challenge from a denoising perspective by introducing SparGCP, which incorporates graph sparsification and a conformal prediction-specific objective into GNN training. SparGCP employs a parameterized graph sparsification module to filter out task-irrelevant edges, thereby improving conformal prediction efficiency. Extensive experiments on real-world graph datasets demonstrate that SparGCP outperforms existing methods, reducing prediction set sizes by an average of 32\% and scaling seamlessly to large networks on commodity GPUs.