RUB Bauwesen CompEng

New member for the CompEng Lecturer Team: Interview with Prof. Dr. Tobias Glasmachers

We are happy to announce that from the summer semester 2019 Prof. Dr. Tobias Glasmachers will be joining the CompEng lecturer team. The lecture “Machine Learning: Supervised Methods” will be part of the CompEng curriculum.

He received his PhD from the Faculty of Mathematics at the Ruhr-University Bochum while working in Christian Igel's group at the Institute for Neural Computation at the RUB. In 2009 Prof. Dr. Glasmachers left Bochum and took a position as a post doc in Jürgen Schmidhuber's group at IDSIA, Lugano, Switzerland. After 3 years, his way led him back to the Ruhr-University. First as Junior professor for theory of machine learning at the Neural Computation, followed by the promotion to a full professor in 2018 at the same institute. In the following interview Prof. Dr. Glasmachers gives a short inside into his research field and why machine learning is becoming an important field in engineering.


Since 2018 you have been holding a full professorship at the Institute for Neural Computation at the Ruhr-University Bochum. At first glance, this has less to do with engineering science, especially with Computational Engineering. What is your research profile?

My research interests are machine learning and optimization. These two areas are intimately connected: optimization algorithms are the very basis of nearly all training procedures. From a methods perspective, machine learning is indeed rather different from engineering, however, the problems tackled are often very similar.


What does Machine Learning mean?

Machine learning is all about training predictive models from data. These can be full-blown process models, but often they are reduced to simpler tasks like classification. The data-driven approach, in a sense, contrats engineering. Instead of using expert knowledge and human problem understanding, the training algorithm figures out by itself how to build the model, relying only on the training data. Hence, models are built in an automated way, and they automatically improve as more data becomes available. However, in practice this does not mean that the human expert is completely outside the loop. Data scientists select and validate suitable model types and tune their parameters to the task at hand. More often than not, engineering expertise is key to success, as it guides the important process of data cleaning and preparation.


What future prospects does Machine Learning offer in the field of engineering?

Machine learning is becoming increasingly important in engineering, and there is a strong demand from industry for engineers with this competency. It is of particular interest in domains where exact modelling is infeasible, either due to limited knowledge, or due to overwhelming complexity, and where data is available or easy to obtain. Autonomous driving is an excellent example of such a problem domain. Machine learning is a very general tool, and as such it finds applications in nearly all sub-fields of engineering.


What possibilities do students have to further explore the field of machine learning?

The Institute for Neural Computation ( offers a whole series of courses on the topic, for example unsupervised learning and a specialized course on deep learning for computer vision. Nowadays there are also plenty of online courses available. There is excellent and easy-to-use open source machine learning software available for everyone, so getting your hands dirty is easier than ever.