Testing the robustness of attribution methods for convolutional neural networks in MRI-based Alzheimer's disease classification
By Fabian Eitel, Kerstin Ritter. 2019
85 citations recorded by the source corpus.
Attribution methods are an easy to use tool for investigating and validating machine learning models. Multiple methods have been suggested in the literature and it is not yet clear which method is most suitable for a given task. In this study, we tested the robustness of four attribution methods, namely gradient*input, guided backpropagation, layer-wise relevance propagation and occlusion, for the task of Alzheimer's disease classification. We have repeatedly trained a convolutional neural network (CNN) with identical training settings in order to separate structural MRI data of patients with Alzheimer's disease and healthy controls. Afterwards, we produced attribution maps for each subject in the test data and quantitatively compared them across models and attribution methods. We show that visual comparison is not sufficient and that some widely used attribution methods produce highly inconsistent outcomes.
Subfields: Artificial Intelligence, Neurology
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Amitojdeep Singh
- Revant Teotia
- Cuntai Guan
- Carl Vondrick
- Junfeng Yang
- Xin Eric Wang
- Sachit Menon
- Vasudevan Lakshminarayanan
Citations in the corpus
- A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI (2019, background): “Comparison between 4 different visualizations is performed in [157].”
- Doubly Right Object Recognition: A Why Prompt for Visual Rationales (2022, background): “Learning models that can explain their own decision is important for building trustworthy systems, especially in applications that require humanmachine interactions [2, 15, 37, 50].”
- Explainable Deep Learning Models in Medical Image Analysis (2020, background): “It was demonstrated on 3D MNIST and an MRI dataset where it returned images with similar atrophy conditions.”