Toward Simple and Robust Contrastive Explanations for Image Classification by Leveraging Instance Similarity and Concept Relevance
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866916817494081536 |
|---|---|
| author | Kaidashova, Yuliia Finzel, Bettina Schmid, Ute |
| author_facet | Kaidashova, Yuliia Finzel, Bettina Schmid, Ute |
| contents | Understanding why a classification model prefers one class over another for an input instance is the challenge of contrastive explanation. This work implements concept-based contrastive explanations for image classification by leveraging the similarity of instance embeddings and relevance of human-understandable concepts used by a fine-tuned deep learning model. Our approach extracts concepts with their relevance score, computes contrasts for similar instances, and evaluates the resulting contrastive explanations based on explanation complexity. Robustness is tested for different image augmentations. Two research questions are addressed: (1) whether explanation complexity varies across different relevance ranges, and (2) whether explanation complexity remains consistent under image augmentations such as rotation and noise. The results confirm that for our experiments higher concept relevance leads to shorter, less complex explanations, while lower relevance results in longer, more diffuse explanations. Additionally, explanations show varying degrees of robustness. The discussion of these findings offers insights into the potential of building more interpretable and robust AI systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_23975 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Toward Simple and Robust Contrastive Explanations for Image Classification by Leveraging Instance Similarity and Concept Relevance Kaidashova, Yuliia Finzel, Bettina Schmid, Ute Computer Vision and Pattern Recognition 68T07 I.2; I.4 Understanding why a classification model prefers one class over another for an input instance is the challenge of contrastive explanation. This work implements concept-based contrastive explanations for image classification by leveraging the similarity of instance embeddings and relevance of human-understandable concepts used by a fine-tuned deep learning model. Our approach extracts concepts with their relevance score, computes contrasts for similar instances, and evaluates the resulting contrastive explanations based on explanation complexity. Robustness is tested for different image augmentations. Two research questions are addressed: (1) whether explanation complexity varies across different relevance ranges, and (2) whether explanation complexity remains consistent under image augmentations such as rotation and noise. The results confirm that for our experiments higher concept relevance leads to shorter, less complex explanations, while lower relevance results in longer, more diffuse explanations. Additionally, explanations show varying degrees of robustness. The discussion of these findings offers insights into the potential of building more interpretable and robust AI systems. |
| title | Toward Simple and Robust Contrastive Explanations for Image Classification by Leveraging Instance Similarity and Concept Relevance |
| topic | Computer Vision and Pattern Recognition 68T07 I.2; I.4 |
| url | https://arxiv.org/abs/2506.23975 |