Mathematics of Machine Learning

This course focuses on statistical learning theory, covering classification and regression. Concepts include PAC model, empirical risk minimization, and Vapnik-Chervonenkis theory. Linear classification methods like perceptron, logistic regression, and support vector machines, including the kernel trick, will be studied. Additionally, the curriculum delves into feedforward neural networks, training using stochastic optimization, techniques for regularization, and methods for validation and testing of models. By the end, students will have a solid understanding of these topics, enabling them to apply advanced machine learning techniques effectively in classification and regression scenarios.

Learning outcomes / Competencies:

• Demonstrate Understanding: Articulate fundamental mathematical principles underpinning supervised learning and statistical learning theory, showcasing a strong grasp of the theoretical foundations.
• Apply Classification and Regression Algorithms: Acquire proficiency in key algorithms for both classification and nonlinear regression, allowing students to confidently select and employ suitable methods based on problem requirements.
• Implement Algorithms: Implement chosen classification methods effectively using commonly used software tools, translating theoretical knowledge into practical applications.
• Evaluate Results: Critically assess the outcomes of machine learning procedures, identifying strengths, weaknesses, and potential sources of error, thereby enhancing analytical skills in model evaluation.
• Problem Solving: Apply acquired knowledge to real-world scenarios, strategically choosing appropriate classification methods and skillfully interpreting and presenting results, showcasing an aptitude for practical problem-solving in the context of supervised learning.