Astrid Stehl

Title: Astrid Stehl: Inventor in Machine-Learning-Based Document Categorization

Introduction

Astrid Stehl is an inventor associated with Göttingen, Germany, and Sartorius Stedim Data Analytics AB. Her patent work addresses the use of machine learning to categorize documents while communicating how reliable the classification may be. Stehl is credited with 1 patent.

Latest Patents

Stehl’s patent, “Accelerated Document Categorization Using Machine Learning,” describes a computer-implemented method for classifying documents. The method uses a machine-learning model that includes an artificial neural network and an attention mechanism to assign documents to at least two classes.

For each document, the system determines a confidence value based on outputs from one or more neural-network nodes. It then assigns the document to a category associated with a particular degree of classification credibility. The system can display the document alongside its category and attention information, which identifies parts of the document that were significant to the classification.

Career Highlights

Stehl’s listed professional affiliation is Sartorius Stedim Data Analytics AB. Her patent highlights an approach that combines document classification, confidence assessment, and attention-based explanations in a single workflow.

Collaborations

Alexander Sutherland is identified as Stehl’s coworker. The available information does not specify a particular joint patent or project between them.

Conclusion

Astrid Stehl’s patent focuses on making machine-learning-based document categorization more informative by pairing classification results with confidence categories and attention information.