Joschka Braun

Title: Joschka Braun: Inventor in Medical Imaging

Introduction

Joschka Braun is an inventor based in Darmstadt, Germany. His patent work addresses the challenge of improving the quality of labeled data used to train machine-learning systems for medical imaging.

Latest Patents

Braun is associated with 1 patent, titled “Re-weighted self-influence for labeling noise removal in medical imaging data.” The invention describes image-processing techniques that estimate the reliability of labeled input images. It assigns each image a self-influence weight based on its self-influence and a threshold percentage representing the expected proportion of correctly labeled images. A training loss function then uses these weights when evaluating the labeled images, helping account for potential labeling errors in the training data.

Career Highlights

Braun works at Covera Health, Inc. His patent reflects an application of machine learning to medical imaging, with a focus on reducing the impact of noisy labels during model training.

Collaborations

Braun’s coworker Micha Kornreich is also associated with his professional context.

Conclusion

Joschka Braun’s patent focuses on weighting medical-imaging training data according to label reliability, an approach intended to help machine-learning systems learn from more dependable examples.