Divide et Calibra: Multiclass Local Calibration via Vector Quantization
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911701434105856 |
|---|---|
| author | Barbera, Cesare Perini, Lorenzo De Toni, Giovanni Passerini, Andrea Pugnana, Andrea |
| author_facet | Barbera, Cesare Perini, Lorenzo De Toni, Giovanni Passerini, Andrea Pugnana, Andrea |
| contents | Accurate and well-calibrated Machine Learning (ML) models are mandatory in high-stakes settings, yet effective multiclass calibration remains challenging: global approaches assume calibration errors are homogeneous across the latent space, while local methods often rely on latent-space dimensionality reduction, which leads to information loss. To address these issues, we propose a compositional approach to multiclass calibration, where region-specific calibration maps are constructed from shared codeword-dependent factors. We instantiate this idea via Vector Quantization (VQ), which induces a structured partition of the representation space, and an indexed parameterization of Dirichlet concentrations that enables parameter sharing across regions. Our approach learns heterogeneous calibration maps that generalize well even to sparse regions of the latent space. Experiments on benchmark datasets show significant improvements in local calibration while maintaining competitive global calibration and predictive performance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_21060 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Divide et Calibra: Multiclass Local Calibration via Vector Quantization Barbera, Cesare Perini, Lorenzo De Toni, Giovanni Passerini, Andrea Pugnana, Andrea Machine Learning Artificial Intelligence Accurate and well-calibrated Machine Learning (ML) models are mandatory in high-stakes settings, yet effective multiclass calibration remains challenging: global approaches assume calibration errors are homogeneous across the latent space, while local methods often rely on latent-space dimensionality reduction, which leads to information loss. To address these issues, we propose a compositional approach to multiclass calibration, where region-specific calibration maps are constructed from shared codeword-dependent factors. We instantiate this idea via Vector Quantization (VQ), which induces a structured partition of the representation space, and an indexed parameterization of Dirichlet concentrations that enables parameter sharing across regions. Our approach learns heterogeneous calibration maps that generalize well even to sparse regions of the latent space. Experiments on benchmark datasets show significant improvements in local calibration while maintaining competitive global calibration and predictive performance. |
| title | Divide et Calibra: Multiclass Local Calibration via Vector Quantization |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2605.21060 |