1. Classification Slides by Greg Grudic, CSCI 3202 Fall 2007
Modified by Longin Jan Latecki
Greg Grudic
Intro AI

2. Why Classification? Uncertainty world Signals Sensing Actions
Computation
Not typically
addressed in CS
State
Symbols
(The
Grounding
Problem)
Decisions/Planning
Agent
Greg Grudic
Introduction to AI

3. Identifying (and Navigating) Paths
Non-path
Data
Data
Construct a
Classifier
Classifier
Data
Path labeled Image
12/5/2018
Intro AI

4. This Class: Classification Models
Collect Training data
Construct Model: happy = F(feature space)
Make a prediction
High
Dimensional
Feature (input)
Space
Greg Grudic
Intro AI

5. Goal of Classification
Give Training Data
GOAL:
Construct a model
Model Property: Minimum error rate on future (unseen) data:
Greg Grudic
Intro AI

6. Measuring Model Accuracy: Classification
Assume a set of data
Classification accuracy
Where
Greg Grudic
Intro AI

7. Binary Classification
A binary classifier is a mapping from a set of d inputs to a single output which can take on one of TWO values (e.g. path/no path)
In the most general setting
Specifying the output classes as -1 and +1 is arbitrary!
Often done as a mathematical convenience
Greg Grudic
Intro AI

8. A Binary Classifier Given learning data: A model is constructed:
Classification
Model
Not in
learning set!
Greg Grudic
Intro AI

9. Classification Learning Data…
Example 1 1
Example 2
0.4235 -1
Example 3
0.8913
Example 4 …
Greg Grudic
Intro AI

10. The Learning Data
Matrix Representation of N learning examples of d dimensional inputs
Greg Grudic
Intro AI

11. Graphical Representation of 2D Classification Training Data
Greg Grudic
Intro AI

12. Linear Separating Hyper-Planes: Discriminative Classifiers
How many lines can separate these points?
NO!
Greg Grudic
Intro AI

13. Greg Grudic
Intro AI

14. Greg Grudic
Intro AI

15. Greg Grudic
Intro AI

16. Greg Grudic
Intro AI

17. Is this Data Linearly Separable?
NO!
Greg Grudic
Intro AI

18. Is this Data Linearly Separable?
YES!
Greg Grudic
Intro AI

19. Is this Data Linearly Separable?
NO!
Greg Grudic
Intro AI

20. Is this Data Linearly Separable?
YES!
Greg Grudic
Intro AI