Syllabus
1. Data preprocessing.
2. How to compare machine learning algorithms.
3. Methods of supervised learning: classification (decision trees, Bayesian classifiers, logistic regression, discriminant analysis, nearest neighbour, support vector machines, neural networks, combination of classifiers - boosting) and their applications in genomics, proteomics and system biology.
4. Methods of unsupervised learning: clustering (partition clustering, k-means, hierarchical clustering, validation of clustering) and its application in bioinformatics.
5. Probabilistic graphical models (Bayesian networks, Gaussian networks) and their applications (in genomics and system biology).
6. Optimization and its application in bioinformatics. The lecture is accompanied by a seminary, where the methods from the lecture will be applied to real and artificial biological data. For implementing the algorithms there will be used mainly an interactive language Python with libraries for machine learning and processing of biological data. The seminary is completed by student projects.
Annotation
Traditional computer science techniques and algorithms fail to solve complex biological problems. However, machine learning techniques can be applied to analyse and process huge volume of biological data. The lecture presents several areas where machine learning is used to process biological data. The students of the course are supposed to know basics of bioinformatics, which they can learn by passing the course Bioinformatics
Algorithms NTIN084, or some similar course at another school.