Machine Learning Laboratory Exercises

Course: Machine Learning

1. Linear regression

• Simple regression
• Polynomial regression and the influence of noise
• Model selection
• Regularized regression
• L1-regularization and L2-regularization
• Features of different scales
• Multicollinearity

2. Linear discriminative models and logistic regression

• Linear regression as a classifier
• Multi-class classification
• Logistic regression
• Logistic regression analysis
• Regularized logistic regression
• Logistic regression with mapping function

3. Support vector machine and non-parametric methods

• Support Vector Machine (SVM) classifier
• Non-linear SVM
• Optimization of SVM hyperparameters
• Impact of feature standardization in SVM
• The k-nearest neighbor algorithm
• Influence of the hyperparameter k
• Non-essential features

4. Parameter estimation, probabilistic graphic models, grouping

• Maximum likelihood estimation
• Maximum posterior probability estimation
• Analysis of the Iris data set
• Probabilistic graphical models -- Bayesian networks
• Explaining effect
• The k-means algorithm
• Gaussian mixture model
• Grouping evaluation

Credits

© Faculty of Electrical Engineering and Computing, University of Zagreb, 2022/2023

Contributors names and contact info

Author	GitHub	e-mail
Enio Krizman	@kr1zzo	enio.krizman@fer.hr

Academic title	Lecturer
Prof. Dr. Sc.	Jan Šnajder