Dr. Jaume Bacardit
My research interests include the development of machine learning methods for large-scale problems and their application to challenging problems, mostly involving biological data.
I have published papers on algorithmic advances to improve the scalability of machine learning methods, tackling challenges such as large dimensionality spaces, large sets of records, postprocessing operators or the use of data-intensive computing technologies such as GPUs and MapReduce.
The main focus of my applied research on biological data is knowledge discovery: analyzing the structure of the data mining models to discover useful knowledge, such as (panels of) biomarkers or functional networks and in this way bring the data mining process closer to the domain experts. My methods have been applied to a variety of biological/biomedical domains: the proces of germination in plants, cancer in humans or osteoarthitis both in humans and model organisms and multiple data-generating biotechnologies: transcriptomics, proteomics, lipidomics, etc.
Currently I lead the data mining efforts of the D-BOARD FP7 project that has as objective the discovery of novel biomarkers for Osteoarthitis. This project generates data of many different types, and the data mining is central to integrate all this heterogeneous information and distill biomarkers with diagnostic and prognostic power.
My PhD thesis dealt with the Pittburgh model of Learning Classifier Systems (LCS). Specifically, the thesis had the following objectives:
From 2005 to 2007 I worked as RA applying LCSs to protein structure prediction, in an EPSRC project called
Robust Prediction with Explanatory Power for Protein Structure and Related Prediction Problems
Research team and visitors: