# Interpretable and Fine-Grained Visual Explanations for Convolutional Neural Networks

## TL;DR

This paper introduces a post-hoc visual explanation method for CNNs that highlights evidence in images for specific predictions, defending against adversarial artifacts and providing fine-grained, interpretable insights.

## Contribution

It presents a novel gradient filtering technique during optimization that enhances explanation quality without human tuning, improving interpretability and robustness.

## Key findings

- Effective in highlighting relevant image evidence
- Resistant to adversarial artifacts in explanations
- Applicable across multiple models and datasets

## Abstract

To verify and validate networks, it is essential to gain insight into their decisions, limitations as well as possible shortcomings of training data. In this work, we propose a post-hoc, optimization based visual explanation method, which highlights the evidence in the input image for a specific prediction. Our approach is based on a novel technique to defend against adversarial evidence (i.e. faulty evidence due to artefacts) by filtering gradients during optimization. The defense does not depend on human-tuned parameters. It enables explanations which are both fine-grained and preserve the characteristics of images, such as edges and colors. The explanations are interpretable, suited for visualizing detailed evidence and can be tested as they are valid model inputs. We qualitatively and quantitatively evaluate our approach on a multitude of models and datasets.