Lab: Machine Learning in Production Engineering

Through machine learning production and manufacturing systems can be enabled to learn processes and algorithms autonomously and to make situation-based decisions. However, a major challenge in building neural networks is that the network architecture, the learning parameters, the dataset size and the evaluation criteria are problem-specific and, thus, there are no general guidelines for setting up neural networks.

The aim of this lab is to impart the knowledge of how to solve practical and application-oriented problems in production engineering with neural networks. In the process, the participants learn to link their theoretical knowledge of artificial intelligence with the programming of neural networks and receive direct feedback by a real world example in a practical lesson.

The course is divided into two sections. In the first section, the participants should acquire the theoretical knowledge of machine learning and familiarize themselves with the programming of neural networks. In the second section, the students have the task to automate a recycling process for beverage closures using machine learning methods at the institute’s lab.

By the end of this course, the students will know how to:

• generate datasets for the training of neural networks,
• program neural networks for computer vision in python,
• train neural networks on the basis of a dataset,
• evaluate the performance of neural networks,
• implement neural networks for tasks in production engineering,
• assess for witch tasks neural networks are applicable.

Registration via Stud.IP at the beginning of the semester. There is a maximum of 40 participants.