The Manifold Hypothesis for Gradient-Based Explanations
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2022
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866909254554746880 |
|---|---|
| author | Bordt, Sebastian Upadhyay, Uddeshya Akata, Zeynep von Luxburg, Ulrike |
| author_facet | Bordt, Sebastian Upadhyay, Uddeshya Akata, Zeynep von Luxburg, Ulrike |
| contents | When do gradient-based explanation algorithms provide perceptually-aligned explanations? We propose a criterion: the feature attributions need to be aligned with the tangent space of the data manifold. To provide evidence for this hypothesis, we introduce a framework based on variational autoencoders that allows to estimate and generate image manifolds. Through experiments across a range of different datasets -- MNIST, EMNIST, CIFAR10, X-ray pneumonia and Diabetic Retinopathy detection -- we demonstrate that the more a feature attribution is aligned with the tangent space of the data, the more perceptually-aligned it tends to be. We then show that the attributions provided by popular post-hoc methods such as Integrated Gradients and SmoothGrad are more strongly aligned with the data manifold than the raw gradient. Adversarial training also improves the alignment of model gradients with the data manifold. As a consequence, we suggest that explanation algorithms should actively strive to align their explanations with the data manifold. This is an extended version of a CVPR Workshop paper. Code is available at https://github.com/tml-tuebingen/explanations-manifold. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2206_07387 |
| institution | arXiv |
| publishDate | 2022 |
| record_format | arxiv |
| spellingShingle | The Manifold Hypothesis for Gradient-Based Explanations Bordt, Sebastian Upadhyay, Uddeshya Akata, Zeynep von Luxburg, Ulrike Machine Learning Computer Vision and Pattern Recognition When do gradient-based explanation algorithms provide perceptually-aligned explanations? We propose a criterion: the feature attributions need to be aligned with the tangent space of the data manifold. To provide evidence for this hypothesis, we introduce a framework based on variational autoencoders that allows to estimate and generate image manifolds. Through experiments across a range of different datasets -- MNIST, EMNIST, CIFAR10, X-ray pneumonia and Diabetic Retinopathy detection -- we demonstrate that the more a feature attribution is aligned with the tangent space of the data, the more perceptually-aligned it tends to be. We then show that the attributions provided by popular post-hoc methods such as Integrated Gradients and SmoothGrad are more strongly aligned with the data manifold than the raw gradient. Adversarial training also improves the alignment of model gradients with the data manifold. As a consequence, we suggest that explanation algorithms should actively strive to align their explanations with the data manifold. This is an extended version of a CVPR Workshop paper. Code is available at https://github.com/tml-tuebingen/explanations-manifold. |
| title | The Manifold Hypothesis for Gradient-Based Explanations |
| topic | Machine Learning Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2206.07387 |