Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
To unify previously fragmented approaches to uncertainty in graph neural networks, this work models node embeddings as random graph signals via a doubly-spectral stochastic expansion combining graph Fourier filters with orthogonal-polynomial chaos coordinates. This yields a single representation with provable approximation guarantees whose distinct spectral coefficients simultaneously drive energy-based OOD scoring, logit variance estimation, and quadrature-based calibrated predictions. Standalone DSS-GNN achieves the lowest Brier score across 14 node classification benchmarks without post-hoc calibration, while a residual hybrid variant delivers state-of-the-art OOD detection and concept-shift robustness across 7 GOOD benchmarks under standard ERM.
Calibration, out-of-distribution detection, and concept shift in graph networks can all be solved by a single representation—coupling graph Fourier transforms with polynomial chaos beats specialized baselines across all three tasks without post-hoc tuning.
Reliable deployment of graph neural networks requires calibration, out-of-distribution (OOD) detection, and robustness to distribution shift, yet existing methods address these needs with separate models and objectives. We model uncertain node embeddings as random graph signals: graph Fourier filters capture structural variation, and a scalar orthogonal-polynomial chaos coordinate captures latent stochastic variation. The resulting doubly-spectral stochastic (DSS) expansion supplies task-matched readouts from one representation: the mean coefficient encodes class evidence for the energy-based OOD score, the higher-order coefficients encode structured logit variation, and quadrature averaging over the chaos coordinate defines the single predictive distribution used for prediction and calibration. A capacity theorem shows that, under a full-rank feature assumption, a restricted subfamily matches the chaos coefficients of any Gaussian-latent random graph signal, with exponentially decaying truncation error under a growth condition; the task-level claims are established empirically. DSS-GNN has two deployment modes: standalone, or as a residual branch beside a deterministic encoder (DSS-Hybrid). Standalone DSS-GNN achieves the lowest Brier score among the compared uncertainty-aware baselines on all 14 node classification benchmarks without post-hoc correction; DSS-Hybrid achieves the best AUROC on most node-OOD settings, competitive cross-graph OOD detection, and the strongest shifted accuracy on all 7 GOOD concept-shift benchmarks under standard empirical risk minimization (ERM). Cross-evaluating both modes on all three tasks shows that each remains effective on the other's tasks, with documented exceptions, and yields explicit deployment guidance.