My research focus lies in the field of statistical machine learning. Machine learning enables IT systems to recognize patterns and laws on the basis of existing data and algorithms and to develop solutions. Artificial knowledge is generated from experience. The knowledge gained from the data can be generalised and used for new problem solutions or for the analysis of previously unknown data.
In particular, I focus on how this can be done efficiently with core based methods in both direct and inverse learning problems with very large amounts of data ("big data"). The development in the field of Big Data technology has also given machine learning a great boost. I have mainly dealt with distributed learning, stochastic approximation methods and subsampling methods.
Also used are so-called neuronal networks, which function according to the model of the human brain. My research in this field was stimulated here at ISA. Designing neural network architectures to solve problems is complex. Numerous hyper parameters to be determined and many loss functions that can be selected for optimization make the matter even more challenging. With my research I would like to contribute to a better theoretical understanding of these complex systems.
Why is this research direction so interesting / forward-looking?
Extensive research activities aim at neural network architectures for autonomous learning. Developments in the field of autonomous driving are hardly conceivable without machine learning. The level of intelligence that can be achieved with deep learning is necessary for autonomous vehicles to recognize and interpret their environment correctly.
A second important area of application is optimised energy management. Climate change and energy system transformation are among the greatest challenges facing politics, society and industry today. Data science methods such as machine learning make an increasingly complex energy market manageable. To ensure that demand can always be optimally met, it is necessary to keep a close eye on both the framework conditions for energy generation and the expected consumption.
Machine Learning is the ideal solution for this task, in which knowledge must also be derived from experience. Machine learning algorithms help to match demand and supply or to detect anomalies in power consumption.
Deep learning can also be used particularly effectively in medical image processing, since high-quality data is available and convolutional neural networks can classify images. Deep learning systems are at least as effective at classifying skin cancer diseases as dermatologists.
You and Mr. Florian Dumpert are organising the "Stuttgart Workshop on Statistical Learning" in July. How important is the exchange between PhD students and postdocs in your field of research?
Basically, I find the exchange of PhD students and postdocs among each other important, not only in my field of research. The idea for my workshop arose out of my personal "need". There are numerous offers for workshops that introduce the topic of Machine Learning, but are not well adapted for more advanced researchers. There are also workshops organized by professors where PhD students and postdocs typically have little or no room to talk about their own work. With my workshop I try to close this gap.
Mrs. Mücke, you also did research at the Universita' degli Studi di Genova. What is the difference to research and teaching at the University of Stuttgart or Potsdam?
In Genoa I have experienced that there are (very) different ways to research meaningfully; different cultural backgrounds are enriching. The colleagues in Genoa were very communicative and there was always an interested and lively exchange about current research topics. Compared to my experiences at the University of Stuttgart, Genoa and Potsdam had less strict and more permeable hierarchies. I really appreciate this openness. Thus it was possible to work "at eye level" with each other, both in research and in teaching.
However, I have also noticed how privileged I am as a young scientist at a German university. Both in Potsdam and in Stuttgart, for example, there is an effort to stand up for the special needs of female scientists, and in particular mothers in research, which is represented by an Equal Opportunities Unit. I am grateful for this. The statistics show that such institutions are important: The proportion of women in science nationwide is only 28%. This puts Germany at the bottom of the international rankings, only in 38th place.In Stuttgart, the proportion of female professors in the mathematics department is just 16%. This confirms the need to support young female scientists, especially mothers. However, there is still room for improvement in practice in this area. In general, the situation for women habilitated in this department is still very difficult when it comes to designing and running their own courses. I hope that something will change in this respect. This would make the University of Stuttgart more attractive.
Dr. Nicole Mücke
Institute of Stochastics and Applications