{"slug": "residual-calibration-via-local-feature-space-refinement", "title": "Residual Calibration via Local Feature-Space Refinement", "summary": "Researchers posting arXiv paper 2610.30652v1 introduced Local Adaptive Probability Estimation (LAPE), a post-hoc residual calibration framework that treats a global calibrator's output as a stable anchor and refines individual predictions using non-parametric local evidence from neighboring calibration samples. The authors also extend LAPE to multi-view inputs through Cascade Feature Augmentation Fusion (CFAF), and report that experiments show the methods improve calibration quality over various existing baselines while remaining efficient and robust across settings. The work targets spatially varying miscalibration in local regions of the learned feature space that score-based post-hoc transformations often miss.", "body_md": "arXiv:2610.30652v1 Announce Type: new \nAbstract: Deep neural networks often struggle to provide reliable confidence estimates, making post-hoc calibration essential for reliable decision-making. Existing post-hoc methods largely rely on score-based transformations, which improve marginal calibration but often fail to capture spatially varying miscalibration in local regions of the learned feature space. To address this gap, we propose Local Adaptive Probability Estimation (LAPE), a post-hoc residual calibration framework. Specifically, LAPE treats the global calibrator output as a stable anchor and refines the individual prediction by leveraging non-parametric local evidence from neighboring calibration samples. As a plug-and-play refinement module, LAPE effectively utilizes the information about feature-space geometry and can reuse the calibration data that are used to construct the global calibrator. Furthermore, we extend LAPE to the setting with multi-view inputs through Cascade Feature Augmentation Fusion (CFAF). Extensive experiments demonstrate that the proposed methods can improve the calibration quality of various existing baselines and remain efficient and robust across different settings.", "url": "https://wpnews.pro/news/residual-calibration-via-local-feature-space-refinement", "canonical_source": "https://arxiv.org/abs/2610.30652", "published_at": "2026-10-23 04:00:00+00:00", "updated_at": "2026-10-23 04:13:49.461903+00:00", "lang": "en", "topics": ["machine-learning", "neural-networks", "ai-research", "artificial-intelligence"], "entities": ["Local Adaptive Probability Estimation (LAPE)", "Cascade Feature Augmentation Fusion (CFAF)", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/residual-calibration-via-local-feature-space-refinement", "markdown": "https://wpnews.pro/news/residual-calibration-via-local-feature-space-refinement.md", "text": "https://wpnews.pro/news/residual-calibration-via-local-feature-space-refinement.txt", "jsonld": "https://wpnews.pro/news/residual-calibration-via-local-feature-space-refinement.jsonld"}}