Explainable AI

Deep learning models are often regarded as black boxes, but this limits the ability to trust the decisions they make. Thus, methods are needed to enable understanding of the mechanisms behind the decision making process.

Home  /  Harmonisation  /  Segmentation  /  Translation  /  Privacy  /  Explainable AI

Papers

In this work we present ExPeRT: an explainable prototype-based model specifically designed for regression tasks.

We trained convolutional neural nets to differentiate patients with dementia from healthy controls, and applied layerwise relevance propagation to procure individual-level explanations of the model predictions. Through extensive validations we demonstrate that patterns recognized by the model corroborate existing knowledge of neuropathology in dementia.

Review article, explores: data availability, interpretability, model bias and data privacy.

Development of age prediction model using data from the UK Biobank, and exploration of correlations with UK Biobank Variables. We also explore the effect of registration on the model.