The course aims to provide a good understanding how logic-based methods can be applied in the construction of knowledge-based systems within Life Sciences. The flood of large and very large data sets such as gene expressions, molecular interactions and taxonomies requires efficient handling. Specifically, the course leads to an advanced understanding of how learning methods can be applied to solve several bioinformatics problems.
On completion of the course, the student should be able to
use and describe definitions and mathematical notation for information and decision systems, rough sets and rule systems
use other methods for machine learning such as clustering, decision trees, and relate them to rough sets
apply knowledge of rule-based systems and Monte Carlo-based selection methods to formulate and solve classification problems in Life Sciences
Introduction to Boolean functions. Transformation and simplification of Boolean expressions. Information, decision systems and rough sets. Features and their synthesis and selection. Training and validation of models. Statistical properties of models. Examples of applications inLifeSciences include: classification of expressions , prediction of gene functions from time profiles and genomic databases, modelling of transcriptional mechanisms, ligand receptor bindings, drug resistance, prediction of protein function from structure and modelling with clinical and genomic data. Lectures are interspaced with computer labs using real and synthetic data. Ontologies. Machine learning: clustering, rough sets, decision trees, Monte Carlo-based selection, statistical model validity and significance.
Lectures, computer exercises, project and problem solving exercises.
Written closed-book exam at the end of the course (3 credits). Written and computer exercises (1 credit).Project (1 credit).