Machine learning in natural sciences
The Interdisciplinary Research Platform promotes pilot research projects that make use of machine learning and information technology to bridge the gaps between disciplines in the natural sciences. Support and visibility for innovative applications in this area will strengthen the collaborations among diverse research groups and institutes of the Radboud Faculty of Science.
Advances in modern science are impossible to imagine without the use of some sort of computational modelling, probabilistic analysis and large amounts of experimental data crunching. Therefore, it comes as no surprise that machine learning and information technology have become essential tools for several -if not all- research groups in Radboud University and specially in the Faculty of Science.
Particle and astrophysics
In these fields, data-driven knowledge discovery is used for the analysis of models with many parameters and the management huge amounts of data. Diverse tasks such as model selection and evidence computations are performed with the help of tools like deep learning, Monte Carlo sampling and non-convex optimisation.
Machine learning constitutes a cornerstone both at a fundamental and applied level. Describing brain processes from the molecular to the functional and behavioural layer requires ample knowledge of statistical mechanics, theory of complex networks, stochastic processes and many day-to-day ML tools such as deep neural networks, time series analysis and Bayesian modelling.
Molecules and materials
Exciting applications of ML include the use of tensor networks for quantum dynamics, deep learning descriptions of the ground state of many body systems, Monte Carlo simulations in theoretical chemistry, among others. Moreover, in a remarkable bidirectional fertilisation between experiment and theory, researchers are trying to produce hardware that performs neuromorphic computations at atomic scale.
Biology and Life sciences
Researchers of the Faculty of Science have access to very detailed measurements technologies. Remarkably, sequencing facilities and molecular spectroscopy enable the collection of data about the genetic expression, transcriptome, proteome and metabolic states with outstanding resolution. Bioinformatics, data science and statistical modelling are crucial to fully profit from this information and uncover the hidden mechanisms of life and diseases.
Computer Science and machine learning groups
Fundamental research in machine learning is also a topic in the Faculty of Science. Significant contributions have been made in the field of inference in large networks, causal discovery, stochastic control theory and quantum inference, among others. There are ongoing applications of advanced statistical mechanics concepts to inference in medical data.