Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2505.22803 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912400758800384 |
|---|---|
| author | Mendes, Pedro Romano, Paolo Garlan, David |
| author_facet | Mendes, Pedro Romano, Paolo Garlan, David |
| contents | Reliable uncertainty estimation is critical for deploying neural networks (NNs) in real-world applications. While existing calibration techniques often rely on post-hoc adjustments or coarse-grained binning methods, they remain limited in scalability, differentiability, and generalization across domains. In this work, we introduce CLUE (Calibration via Learning Uncertainty-Error Alignment), a novel approach that explicitly aligns predicted uncertainty with observed error during training, grounded in the principle that well-calibrated models should produce uncertainty estimates that match their empirical loss. CLUE adopts a novel loss function that jointly optimizes predictive performance and calibration, using summary statistics of uncertainty and loss as proxies. The proposed method is fully differentiable, domain-agnostic, and compatible with standard training pipelines. Through extensive experiments on vision, regression, and language modeling tasks, including out-of-distribution and domain-shift scenarios, we demonstrate that CLUE achieves superior calibration quality and competitive predictive performance with respect to state-of-the-art approaches without imposing significant computational overhead. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_22803 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CLUE: Neural Networks Calibration via Learning Uncertainty-Error alignment Mendes, Pedro Romano, Paolo Garlan, David Machine Learning Artificial Intelligence Reliable uncertainty estimation is critical for deploying neural networks (NNs) in real-world applications. While existing calibration techniques often rely on post-hoc adjustments or coarse-grained binning methods, they remain limited in scalability, differentiability, and generalization across domains. In this work, we introduce CLUE (Calibration via Learning Uncertainty-Error Alignment), a novel approach that explicitly aligns predicted uncertainty with observed error during training, grounded in the principle that well-calibrated models should produce uncertainty estimates that match their empirical loss. CLUE adopts a novel loss function that jointly optimizes predictive performance and calibration, using summary statistics of uncertainty and loss as proxies. The proposed method is fully differentiable, domain-agnostic, and compatible with standard training pipelines. Through extensive experiments on vision, regression, and language modeling tasks, including out-of-distribution and domain-shift scenarios, we demonstrate that CLUE achieves superior calibration quality and competitive predictive performance with respect to state-of-the-art approaches without imposing significant computational overhead. |
| title | CLUE: Neural Networks Calibration via Learning Uncertainty-Error alignment |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2505.22803 |