Learning Systems for Molecular Data Analysis

6 credit points

Syllabus, D-level, 1MB452

This course has been discontinued.

Code: 1MB452
Level: D
Subject(s): Biology
Grading system: Pass with distinction (5), Pass with credit (4), Pass (3), Fail (U)
Finalised: 2 June 2000
Responsible department: Department of Engineering Sciences

Entry requirements

Algebra and geometry. Calculus of several variables. Mathematical statistics. Numerical analysis. Molecular bioinformatics. Gene function.

The course should yield basic knowledge and

understanding for how different types of learning

systems can be used for problems in molecular data analysis.

The course should also

yield a basic understanding of how different types of

learning systems can be used to solve bioinformatic

problems.

Mathematical foundations for learning systems. Basic

concepts such as Bayesian modelling and parametric and

non-parameteric methods, discriminant functions,

decision surfaces, parameter estimation, supervised and

unsupervised learning, interactive learning systems.

generalisation, clustering. Families of methods

like decision trees, artificial neural networks,

Bayesian learning, and prototype based learning,

learning systems for sequence analysis, dynamic

programming, the EM algorithm, Markov chain monte

carlo methods, simulated annealing, evolutionary and

genetic algorithms, hidden Markov chains, phylognetic trees,

stochastic grammars. Everything integrated with applications

in molecular data analysis.

Practicals: Computer practical with real and synthetic data.

Lectures, seminars and laboratory work.

Written or oral examination at the end of the course,

homeworks. The practical part of the course (including tutorials)

corresponds to 2 points (credits).