Search papers, labs, and topics across Lattice.
This paper introduces structure-aware graph abstention for selective forecasting, addressing limitations of existing whole-forecast scoring methods that overlook inter-variable dependencies. By treating instance-level plausibility and relational consistency as distinct reliability axes, the authors operationalize structural consistency using a learned sparse graph and a Dirichlet-style structural energy term. Experimental results on seven long-horizon benchmarks demonstrate that this structural gating often reduces selective MSE compared to prior methods at matched coverage, particularly when cross-variable structure is informative.
A learned graph structure and structural energy term significantly improve selective forecasting reliability by explicitly penalizing violations of inter-variable dependencies, outperforming whole-forecast scoring methods on key benchmarks.
Selective forecasting abstains on high-risk test windows under a retained-coverage budget. Existing gates such as TEM (Brusokas et al., 2025) score each forecast as a whole; for multivariate outputs, trajectories can look plausible while violating dependencies among variables. We treat instance-level plausibility and relational consistency as distinct reliability axes and operationalize the latter via a learned sparse graph and a Dirichlet-style structural energy E_struct, trained with error-weighted graph regularization and score-error alignment. On seven long-horizon benchmarks and four backbones, structural gating often reduces selective MSE versus TEM at matched coverage, with the largest gains where cross-variable structure appears more informative in our benchmarks; gains are not universal, indicating a complementary abstention signal. Table 1 is a Protocol A ranking diagnostic (seed 2024); three-seed deployable Protocol B on an aligned subset is in Table 3 (full validation-to-test grids: Appendix A).
