Lecture 1
Preparation
Read Chapter 1 in the text.
Notes
An Simple Pattern Recognition Example
- Problem Statement
Can the success of Freshman students be predicted from data available
at registration time?
- Pattern Space
- What are the classes?
- What are the measurements available?
- What is the size of the pattern space?
- What are some possible classifier architecures?
- Decision space
- Decision criteria
- The structure of the decision algorithm
- Learning
- How big a sample is needed?
- Error rates
- Adapting
- When is adaption necessary/wise?
- Adaptive methods
Homework
Given the following problem, describe the pattern space, a responable
measurement space, a decision space and a classifier architecure.
How would you find the apparent error rate? What type of adaption is
appropriate here, if any? There is no right answer here, be creative.
A mail order sports clothing outlet wants to classify customers as
likely, possibly or not likely for the purposes of sending catalogs.