Skip to main content Skip to main navigation

Best Paper Award CD-MAKE 2021

| Health & Medicine | Human-Machine Interaction | Awards | Interactive Machine Learning | Cognitive Assistants | Saarbrücken | Osnabrück / Oldenburg

At the International IFIP Cross Domain (CD) Conference for Machine Learning & Knowledge Extraction (MAKE), which took place from 17 to 20 August 2021 virtually, the scientists Dr. Fabrizio Nunnari (from the DFKI research department Cognitive Assistants), Md Abdul Kadir and Prof. Dr.-Ing. Daniel Sonntag (both from the DFKI research department Interactive Machine Learning) received the Best Paper Award for their publication „On the overlap between Grad-CAM saliency maps and explainable visual features in skin cancer images“.

© CD-MAKE 2021

The paper approaches the comparison of features of melanoma identified by dermatologists on the basis of images with features classified by Convolutional Neural Networks (CNNs). XAI features (explainable AI), so-called saliency maps, are a result of the classification. They serve the purpose of checking if the dermatologist's opinion is in accordance with the opinion of the Machine Learning System and if a possible discrepancy can be explained visually. Our publication is the first one to examine this question systematically with different neural networks on the basis of a specific medical use case.

Read more on  CD-MAKE