معرفی
Julia Grabinski is a Researcher and PhD candidate at the Chair for Machine Learning at the University of Mannheim, affiliated with the Data and Web Science Group within the School of Business Informatics and Mathematics. She is supervised by Prof. Dr.-Ing Margret Keuper. Her research focuses on Signal Processing in modern Computer Vision and the Robustness and Reliability of Deep Learning Models.
Her work investigates adversarial robustness, CNN vulnerabilities, and methods to enhance model reliability through techniques such as FrequencyLowCut Pooling and addressing spectral artifacts. She has contributed to understanding how signal preservation and aliasing affect CNN robustness, proposing solutions like spectral artifact-free pooling layers.
Recent publications explore class-wise robustness analysis, improving feature stability during upsampling, and transformer vulnerability in image restoration. These efforts highlight her focus on foundational challenges in deep learning and computer vision.
Julia is part of the Data and Web Science Group, collaborating on advancing robust machine learning models with practical applications in image processing and neural network architectures.
