Guarantee Regions for Local Explanations
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914686015897600 |
|---|---|
| author | Havasi, Marton Parbhoo, Sonali Doshi-Velez, Finale |
| author_facet | Havasi, Marton Parbhoo, Sonali Doshi-Velez, Finale |
| contents | Interpretability methods that utilise local surrogate models (e.g. LIME) are very good at describing the behaviour of the predictive model at a point of interest, but they are not guaranteed to extrapolate to the local region surrounding the point. However, overfitting to the local curvature of the predictive model and malicious tampering can significantly limit extrapolation. We propose an anchor-based algorithm for identifying regions in which local explanations are guaranteed to be correct by explicitly describing those intervals along which the input features can be trusted. Our method produces an interpretable feature-aligned box where the prediction of the local surrogate model is guaranteed to match the predictive model. We demonstrate that our algorithm can be used to find explanations with larger guarantee regions that better cover the data manifold compared to existing baselines. We also show how our method can identify misleading local explanations with significantly poorer guarantee regions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_12737 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Guarantee Regions for Local Explanations Havasi, Marton Parbhoo, Sonali Doshi-Velez, Finale Machine Learning Interpretability methods that utilise local surrogate models (e.g. LIME) are very good at describing the behaviour of the predictive model at a point of interest, but they are not guaranteed to extrapolate to the local region surrounding the point. However, overfitting to the local curvature of the predictive model and malicious tampering can significantly limit extrapolation. We propose an anchor-based algorithm for identifying regions in which local explanations are guaranteed to be correct by explicitly describing those intervals along which the input features can be trusted. Our method produces an interpretable feature-aligned box where the prediction of the local surrogate model is guaranteed to match the predictive model. We demonstrate that our algorithm can be used to find explanations with larger guarantee regions that better cover the data manifold compared to existing baselines. We also show how our method can identify misleading local explanations with significantly poorer guarantee regions. |
| title | Guarantee Regions for Local Explanations |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2402.12737 |