This course introduces key machine learning concepts and is accompanied by tutorials and exercises where machine learning methods are applied to actual bioinformatics problems. After a short recap of probability theory, we introduce probabilistic methods for classification and sequence analysis (Naive Bayes, Mixture Models, Hidden Markov Models). We discuss Expectation Maximization (EM) from a probabilistic perspective and use it for sequence analysis. Linear and logistic regression serve as an entry point to more complex machine learning methods, including kernel methods and neural networks. The lecture covers multiple neural network architectures (CNNs, GNN, Transformers) that are currently used in the bioinformatics community and other research domains. During the tutorials and as part of homework assignments, selected machine learning models are implemented in Python using scikit-learn and pytorch. The course should enable students to understand all common machine learning techniques and devise state of the art classification strategies that can then be applied to problems in bioinformatics and related fields.

Contents:
- Naive Bayes
- Clustering and Mixture Models
- Hidden Markov Models
- Regression and Partial Least Squares
- Kernel Methods
- Neural Networks and Architectures
- Regularization and Model Selection

 

Requirements:
- Linear algebra (basic vector and matrix algebra)
- Analysis (mathematical optimization, Lagrange)
- Programming in Python -- including object oriented programming
- A basic understanding or keen interest in molecular biology and bioinformatics applications