Data Management, Mining and Analysis
Data Management, Mining and Analysis studies methods and applications involving large data volumes or complex data collections in all information systems. This field is growing in importance as more and more data sources are becoming available from scientific instruments, social networks, open data, monitoring systems, sensor networks, etc.
Our research in data management concerns indexing, efficient access, as well as query processing and similarity search. Our solutions typically consist of algorithmic strategies that allow efficient access to relevant data, often with provable correctness guarantees. A notable focus of our research is the use of modern hardware, i.e., multi-core CPUs and graphics cards (GPUs) on standard desktop or laptop computers.
In data analysis, we are interested in concepts and methods for learning from data, featuring both supervised and unsupervised methods. In particular, we contribute clustering methods for high-dimensional and noisy data, as well as unsupervised outlier detection solutions. Supervised approaches include artificial neural networks, in particular for handling text data (Natural Language Processing, NLP) and image data (e.g. remote sensing), where transfer learning plays an important role.
Methodologically, we combine formal problem definitions and analytical discussion with prototype implementations and empirical studies. We use benchmarking data as well as cooperations with domain experts to evaluation the quality, efficiency and scalability of our approaches.