Machine Learning

(3 Credit Hours) · Undergraduate

This course introduces a set of the most popular methods for addressing four central problems in machine learning: dimension reduction, regression, classification, and clustering, with the greatest emphasis on classification. Dimension reduction aims to reduce the dimensions of high-dimensional data to make it easier to visualize and analyze. Regression involves predicting a continuous variable, classification concerns predicting a categorical variable, and clustering seeks to divide data into useful or meaningful groups. The topics likely to be covered in the course include the following, although we may need to omit a few each year due to limited lecture time: principal component analysis, multidimensional scaling, tSNE, UMAP, k-means clustering, hierarchical clustering, nearest neighbor classifiers, linear/quadratic discriminant analysis, Naive Bayes, decision trees, and ensemble methods (bagging, random forest, boosting), as well as artificial neural networks. Deep learning will also be seriously discussed in this course, although, due to limited lecture time, we may focus only on a small, carefully selected set of topics, such as self-supervised learning, transformers, BERT, and GPT.