Since I learned about neural networks in a bachelor seminar, machine learning has always been the focus of my studies. A cooperative doctorate in this field is the right continuation of this path for me. Since I completed both degrees here at Trier University of Applied Sciences and really appreciate the learning atmosphere here, it was clear for me to do my PhD here. In the future, I would like to continue to work with neural networks and do research in this area.
The knowledge gained about the reconstruction properties of autoencoders should either improve an existing application or lead to a new application.
Possible fields of application are, for example, an increase in data protection regarding the training data, anomaly detection or object recognition.
My topic is studying the reconstruction properties of autoencoders.
When autoencoders are trained in a certain way, they become associative memories.
A part of the training data then becomes attractors to which all data in the corresponding basins of attraction converges when the model is applied iteratively. The goal is to explore patterns in convergence behaviour as new data are entered and to use the findings for a concrete application.
My approach is experimental rather than theoretical. A typical workday for me therefore consists of designing suitable experiments, implementing them in Python, running them, and logging them after completion. Often, I also do research for suitable literature.
A typical working day for me also includes sports.
You are leaving the official website of Trier University of Applied Sciences