CORE-STACK+: Meta-Learning for Deep Stacked Generalization

Stacking heterogeneous vision backbones (CNNs, ViTs, and hybrids) is the de facto recipe for accuracy, calibration, and robustness, yet two coupled pathologies limit its returns. Prediction-space multicollinearity ill-conditions the meta-learner's Gram matrix, inflating weight variance and producing brittle solutions on a thin manifold. Calibration collapse compounds constituent miscalibration through naive linear stacking, so adding more models can hurt expected calibration error (ECE). Existing remedies, ridge regularization, greedy selection, model soups, and SWAG address at most one of these issues, and none jointly target conditioning and calibration in heterogeneous prediction pools. We introduce CORE-STACK+, a preconditioning pipeline with four components: (i) a kernelized redundancy filter that removes non-linear inter-model dependencies invisible to Pearson correlation, using Centered Kernel Alignment (CKA) [23]; (ii) a <15K-parameter differentiable meta-feature gate that learns per-sample attention over ensemble statistics; (iii) a spectrum-adaptive Ridge penalty lambdastar=lmax(Chat)/SNR(Chat) derived from a Marchenko-Pastur signal-noise decomposition, eliminating nested cross-validation; and (iv) a Laplace-approximate Bayesian blender replacing inverse-RMSE heuristics. We prove a PAC-Bayes excess-risk bound that, for the first time, jointly accounts for prediction-space redundancy and meta-learner capacity. Across six benchmarks, CORE-STACK+ delivers +1.8% top-1 on ImageNet-1K, -4.2 mCE on ImageNet-C, +0.9 mIoU on ADE20K, and +1.3 AP on COCO, while reducing retained models by 35-57% and inference FLOPs by up to 41%. ECE improves 2.1× over deep ensembles without post hoc temperature scaling.