CS 445/545 Machine Learning

Credit Hours:

4/3

Course Coordinator:

Ehsan Aryafar and Banafsheh Rekabdar

Course Description:

Provides a broad introduction to techniques for building computer systems that learn from experience; conceptual grounding and practical experience with several learning systems; and grounding for advanced study in statistical learning methods, and for work with adaptive technologies used in speech and image processing, robotic planning and control, diagnostic systems, complex system modeling, and iterative optimization. Students gain practical experience implementing and evaluating systems applied to pattern recognition, prediction, and optimization problems.

Prerequisites:

Mth 261 or Mth 343; and CS 302.

Goals:

• Introduce students to several prominent areas of machine learning, including computational learning theory, support vector machines, Bayesian learning and Bayesian networks, and unsupervised learning, and illustrate what types of problems the different methods are suited for.
• Give students hands-on experience with these methods and tools for implementing and using them on real-world problems.
• Give students experience with performing simulations and doing statistical data analysis of the results.
• Provide students with experience in reading and writing summaries of research papers and giving presentations.

Upon the successful completion of this class, students will be able to:

1. Describe the main components of a machine learning system and the major classes of approaches to machine learning.
2. Describe the overall algorithms and special techniques for several machine learning methods, including support vector machines, Bayesian learning, and unsupervised learning, as well as methods for dimensionality reduction.
3. Explain the relative advantages and disadvantages of each of these methods, and list several potential areas of application for these methods.
4. Design training sets and testing sets for machine learning tasks.
5. Use several public domain machine learning tools.
6. Design and run experiments that test the effectiveness of each of the methods listed above and write up the results of such experiments.

Major Topics:

• Supervised Classification and Regression
• Evaluating Classifiers
• Computational Learning Theory
• Support Vector Machines
• Bayesian Learning and Bayesian Networks
• Clustering
• Mixture Models
• Expectation Maximization
• Principal Components Analysis
• Independent Components Analysis
• Multidimensional Scaling

Oral and Written Communications:

Every student is required to submit at least 1 written report (not including exams, tests, quizzes, or commented programs) of typically 2 pages and to make 1 oral presentation of typically 15 minutes duration.

Theoretical Content:

Computational learning theory, probability and statistics as related to machine learning methods (5 lectures).

Solution Design:

Students are required to design solutions to several machine learning problems.