A demo course that illustrates how one can set up a new course, with a custom learning trajectory, by recombining elements of existing courses. Two courses are used here: Course: AI School (UNIZA) | moodle and Course: Machine Learning (UNIZA, Michal Gregor) | moodle.
An introductory Machine Learning course (as last taught in my 2023 class).
Topics include: Introduction and key concepts, local and global generalization, simple machine learning methods (KNN, decision trees, naïve Bayes, ...), intro to data analysis, cluster analysis, supervised learning and optimization (linear regression, logistic regression, gradient descent), evaluation, regularization, artificial neural networks and automatic differentiation, deep learning, deep learning for sequential data, dimensionality reduction, embeddings and face clustering, reinforcement learning, Gaussian processes and Bayesian hyperparameter optimization.
The student should be familiar with the fundamentals of linear algebra and mathematical analysis. They should have intermediate programming skills (and ideally some prior experience with Python).
The student can explain basic concepts from the field of machine learning such as machine learning, implicit and explicit knowledge representation, local and global generalization, underfitting, overfitting, bias, variance, regularization and others. The student is able to explain the principle of the most common machine learning methods. The student can evaluate where and how different machine learning methods can be applied. The student can create applications of machine learning methods and approaches. The student can understand scientific text and, working in a team, perform a literature review and produce a technical report. The student is able to present the results of their work.
The material is for a 6 ECTS credit course with around 162-hour workload with: 2h*13+1h*13 (face-to-face lecture + optional office hours); 30h (individual work on practical exercises); 33h (semester project); 60h (studying, reading, ...).