Sebastian Doerrich
PulseAugur coverage of Sebastian Doerrich — every cluster mentioning Sebastian Doerrich across labs, papers, and developer communities, ranked by signal.
-
Colorist: Statistical Color Matching Enhances Medical Image Classifier Robustness
Researchers have developed a new data augmentation strategy called Colorist, which uses statistical color matching to improve the robustness of medical image classifiers. Unlike deep generative models, Colorist is compu…
-
AI polyp detection models fail in real-world tests due to dataset bias
A new research paper introduces TRUE-Colon, a benchmarking protocol designed to expose the performance gap between AI models trained on curated medical datasets and their performance in real-world clinical settings. The…
-
MoPET method uses mixture-of-experts for medical image classification
Researchers have developed MoPET, a novel parameter-efficient fine-tuning (PEFT) method that utilizes a mixture-of-experts (MoE) approach for medical image classification. MoPET addresses the issue of negative transfer …