AI & Data science

In complex technical and organisational systems data are abundant, but understanding is scarce. Our AI and data-science research therefore develops methods that turn sensor streams, event logs, texts and open data into interpretable models – combining clustering, classification, regression, fuzzy and deep models with domain knowledge – and validates them on real industrial, educational and regional problems.

Our concept is a closed loop: data → representation → learning → interpretation → decision support, in which the use of the results generates new data and new questions. Our newest line of work studies LLM-based agents and the monitoring of human–AI agent systems, with explicit attention to uncertainty, provenance and human oversight. The topics below are the building blocks of this loop; each topic page lists the related publications, newest first.

AI & LLM – large language models in research and decision support, and the economics of LLM services

Biclustering – simultaneous clustering of rows and columns to find local patterns in data matrices

Clustering – fuzzy, graph-based and goal-oriented clustering for segmentation and monitoring

Classification – machine-learning classifiers, software sensors and interpretable classification models

Deep learning – neural network models and explainable AI

Frequent sequence, itemset and association rule mining – pattern mining in event, log and text data

Fuzzy modeling – fuzzy identification and control, interpretable rule-based models

Process mining – discovering and analysing processes from event logs

Ranking – sum of ranking differences and utility-based ranking methods

Reinforcement learning – learning control and decision policies (shared with Optimization)

Regression – regression models and their validation

Spatial analysis – spatial data mining and small-area estimation

Survival analysis – time-to-event modelling

Text mining – network-based analysis of documents

Time series mining – segmentation, similarity and model-based analysis of time series

Visualization – visualisation of high-dimensional data and models