Live-Webinar: Efficient AI Training for Quality Monitoring Using »Active Learning«

Active Learning (AL) is a machine learning approach that significantly reduces the amount of required training data while maintaining or even improving the performance of the AI model. During the learning process, an AI model selects data points with the highest informational value, which are then labeled by a human. This targeted labeling approach greatly reduces manual effort and, consequently, the costs of model development, particularly in areas such as quality monitoring and predictive maintenance.

In this webinar, we will introduce you to the core principles of Active Learning and explore state-of-the-art AL strategies for image and time-series datasets. Using two practical examples – wood defect detection and tool wear detection in milling machines – we will demonstrate how Active Learning can streamline data annotation while simultaneously improving model accuracy.

To conclude, we will present a demonstrator showcasing the entire AL pipeline in practice. The demonstrator iteratively trains AI models from scratch using Active Learning. This workflow highlights how Active Learning reduces labeling effort, accelerates convergence, and demonstrates both efficiency and impact in real-world scenarios.

Referentin: Akshaya Bindu Gowri, Datenanalysesysteme am Fraunhofer IIS in Dresden