Distribution Free Prediction Sets for Node Classification
This addresses the need for reliable uncertainty estimates in GNNs for applications like social networks or biology, though it is incremental as it modifies existing conformal prediction techniques.
The paper tackled the problem of quantifying predictive uncertainty in Graph Neural Networks (GNNs) for node classification by adapting conformal prediction to handle graph-structured data, resulting in tighter and better-calibrated prediction sets compared to naive methods on standard benchmarks.
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.