Metraipolis › archive › dc_c6513df9fdb7

CORE-STACK+: Meta-Learning for Deep Stacked Generalization — arXiv:2609.26905

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.26905
License (as archived)
CC0-1.0
Archived
2026-10-08T10:51:57.059Z
SHA-256 of the archived text
ef52267294e95086cc8e4b8246bbd18f2ccf165dc199fb62fcfc44213908f227
Stable URL of this page
https://metraipolis.com/documento/dc_c6513df9fdb7

How to cite

CORE-STACK+: Meta-Learning for Deep Stacked Generalization — arXiv:2609.26905 — arXiv (en), https://arxiv.org/abs/2609.26905 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c6513df9fdb7 (Metraipolis archive dc_c6513df9fdb7, sha256 ef52267294e95086cc8e4b8246bbd18f2ccf165dc199fb62fcfc44213908f227)
BibTeX
@misc{dcc6513df9fdb7,
  title = {CORE-STACK+: Meta-Learning for Deep Stacked Generalization — arXiv:2609.26905},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.26905},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c6513df9fdb7; sha256 ef52267294e95086cc8e4b8246bbd18f2ccf165dc199fb62fcfc44213908f227},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.26905. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

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 $<15$K-parameter differentiable meta-feature gate that learns per-sample attention over ensemble statistics; (iii) a spectrum-adaptive Ridge penalty $lambda^{star}=lmax(Chat)/SNR(Chat)$ 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 $2.1\times$ over deep ensembles without post hoc temperature scaling.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-24 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG) ; Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.26905 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-08T10:06:16.685Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).