# Toward Simple and Robust Contrastive Explanations for Image Classification by Leveraging Instance Similarity and Concept Relevance

arXiv: https://arxiv.org/abs/2506.23975 · HTML: https://mlsanity.com/paper/2506.23975/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 30 references that are in the map.
- First posted on arXiv 30 Jun 2025.

## Builds on (most cited first, 9 of 9)

- [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://mlsanity.com/paper/1409.1556/) (2014)
- [ImageNet Large Scale Visual Recognition Challenge](https://mlsanity.com/paper/1409.0575/) (2014)
- [Towards A Rigorous Science of Interpretable Machine Learning](https://mlsanity.com/paper/1702.08608/) (2017)
- [Explanation in Artificial Intelligence: Insights from the Social Sciences](https://mlsanity.com/paper/1706.07269/) (2017)
- [Explanations based on the Missing: Towards Contrastive Explanations with Pertinent Negatives](https://mlsanity.com/paper/1802.07623/) (2018)
- [From Attribution Maps to Human-Understandable Explanations through Concept Relevance Propagation](https://mlsanity.com/paper/2206.03208/) (2022)
- [A Comprehensive Taxonomy for Explainable Artificial Intelligence: A Systematic Survey of Surveys on Methods and Concepts](https://mlsanity.com/paper/2105.07190/) (2021)
- [Contrastive Explanations in Neural Networks](https://mlsanity.com/paper/2008.00178/) (2020)
- [Generating Contrastive Explanations for Inductive Logic Programming Based on a Near Miss Approach](https://mlsanity.com/paper/2106.08064/) (2021)

## Abstract (arXiv)

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.