Statistical Machine Learning
Syllabus, Master's level, 1RT700
- Education cycle
- Second cycle
- Main field(s) of study and in-depth level
- Computer Science A1N, Data Science A1N, Image Analysis and Machine Learning A1N, Mathematics A1N, Technology A1N
- Grading system
- Fail (U), Pass (3), Pass with credit (4), Pass with distinction (5)
- Finalised by
- The Faculty Board of Science and Technology, 2 May 2017
- Responsible department
- Department of Information Technology
120 credits including Probability and Statistics, Linear Algebra II, Single Variable Calculus and a course in introductory programming.
Students who pass the course should be able to:
- Structure and divide statistical learning problems into tractable sub-problems, formulate a mathematical solution to the problems and implement this solution using statistical software.
- Use and develop linear and nonlinear models for classification and regression.
- Describe the limitations of linear models and understand how these limitations can be handled using nonlinear models.
- Explain how the quality of a model can be evaluated and how model selection and tuning can be done using cross validation.
- Explain the trade-off between bias and variance.
- Describe the difference between parametric and nonparametric models.
This is an introductory course in statistical machine learning, focusing on classification and regression: linear regression, regularization (ridge regression and LASSO), classification via logistic regression, linear discriminant analysis, classification and regression trees, boosting, neural networks, deep learning; practical considerations such as cross validation, model selection, the bias-variance trade off, applying the methods to real data.
Lectures, problem solving sessions (both with and without computer), laboratory work, feedback on written assignments.
Written exam combined with oral and written presentation of assignments.