Yannic Neuhaus

My research interests are the robustness and explainability of machine learning systems. Currently, I am focused on the detection and mitigation of shortcut learning. This occurs when a machine learning model picks up spurious correlations in the training data possibly resulting in unexpected failure modes after deployment (e.g. computer vision models often rely on spuriously correlated image backgrounds or texture patterns for their predictions instead of the intended visual features).
At the Heinrich Heine University Düsseldorf, I received a B.Sc. in Mathematics and a B.Sc. in Computer Science. Afterwards, I graduated with a M.Sc. in Machine Learning from the University of Tübingen.