Announcement / Posting
CSE 474/574 INTRODUCTION TO MACHINE LEARNING (Fall 2006)
Instructor: Dr. Matthew J. Beal
Time/Place: T R . 12:30-1:50pm . 4 KNOX
Course site: http://www.cse.buffalo.edu/faculty/mbeal/cse574
Credits: 3.00 G / 4.00 UG
Machine Learning aims to build computer systems that can learn from their experiences. The algorithms in Machine Learning are not directly programmed by a person, but instead develop and adapt their own program based on examples of how they should behave.
Machine Learning is an exciting interdisciplinary field with historical roots in computer science, statistics and even philosophy. It draws from and has many applications in a variety of fields including robotics, pattern recognition, computer vision, data compression/coding, natural language processing, physics, and neuroscience, and is rapidly establishing itself as a cornerstone for bioinformatics models and applications.
This course is for those interested in building algorithms that can intelligently represent data and learn for themselves. It is aimed at final-year undergraduates and first-year graduates in CSE, Statistics, and Physics/Math. Some topics include:-
- decision tree learning
- neural networks for supervised learning
- unsupervised learning, such as mixture models and building representations
- probabilistic models in general
- Bayesian belief networks and Bayesian learning, propagation algorithms
- time series models (e.g. Hidden Markov Models and Kalman filters)
- support vector machines
- and reinforcement learning.
This course should leave you with a good knowledge of state-of-the-art machine learning methods, and an understanding of which tools are appropriate for which tasks in real-world applications.
We will use the traditional "Machine Learning" textbook by Mitchell, but also a very recent and excellent book by MacKay called "Information Theory, Inference, and Learning Algorithms".
A basic knowledge of probability and a solid background in calculus and linear algebra will be required (college-level) and will be reviewed in the first week. Most coding will be in MATLAB.
Please visit the course website for more information, including course requirements, textbooks, and grading guidelines.