The CDS laboratory develops novel Machine Learning tools to study cancers, leveraging advanced bioinformatics, computational biology and data science expertise. Our research is interdisciplinary and our group has members trained from classical STEMS to Life Sciences. We collaborate with some of the most advanced experimental facilities and industry partners. Our research is supported by several grants, and we contribute from experimental design to data analysis!

The Machine Learning and Computational Science Laboratory focuses on developing novel machine learning algorithms, particularly in geometric deep learning and neural networks, with applications in computational science. The lab aims to harness geometric deep learning to process complex data and advancing predictive accuracy and pattern recognition in fields like neuroscience, physics, and medicine. It also works on making algorithms interpretable to mitigate biases in AI. Current projects include developing algorithms for quantum information, analyzing neural network biases, studying the visual cortex through neural networks, and creating algorithms for biomedical data analysis.

The Laboratory for Unsupervised Learning and Knowledge Extraction focuses on crafting algorithms to analyze unlabeled or partially labeled data, with applications across physics, chemistry, and medicine. The lab's main research areas include clustering to discover natural data groupings, manifold learning to decode data geometry, and uncovering non-linear correlations in high-dimensional data. Applications span interpreting water physics simulations, designing chemical compounds with desired properties through advanced ML methods like graph neural networks, identifying physical properties in quantum physics and computing, and analyzing neural network vulnerabilities to adversarial attacks.

We develop theory and tools that support formal representation, modelling and analysis of complex stochastic dynamical systems. We are particularly interested in uncovering mechanisms of complex biological systems such as gene regulation, protein signalling and collective behaviour. Following the analogy ’model = computer programme’, we broadly combine the approaches of formal programme verification, mathematical modelling and probabilistic reasoning.

The group is currently in transition from Univ. of Konstanz (Germany).