🔍About

Hi, I'm Coenraad. I am a deep learning researcher focused on understanding the generalization, robustness, and trustworthiness of neural networks. Currently, I am a postdoc at the Intelligent Imaging Lab at the University of Kiel, Germany. Here I work on applications in medical imaging, especially areas related to the diagnosis and prevention of osteoporosis.

Previously, I did a PhD at the North-West University, South Africa, under the supervision of Prof. Marelie Davel and Dr. Marthinus Theunissen. There I primarily worked on developing methods to empirically predict generalization performance from model properties such as margins.

My other areas of interest include:

• Adversarial robustness: Investigating why adversarial examples exist and how to improve models' resilience to adversarial attacks.
• Uncertainty quantification: Exploring methods like Monte Carlo Dropout and Conformal Prediction to provide calibrated confidence estimates for neural network predictions.
• Latent representation analysis: Probing a neural network’s internal representations to gain insights into how it processes information and makes decisions.

If these areas interest you as well, or if you like something I've written, please reach out!