1. 1 Pattern Recognition: Statistical and Neural Lonnie C. Ludeman Lecture 14 Oct 14, 2005 Nanjing University of Science & Technology

2. 2 Lecture 14 Topics 1. Review structures of Optimal Classifier 2. Define Linear functions, hyperplanes, boundaries, unit normals,various distances 3. Use of Linear Discriminant functions for defining classifiers- Examples

3. 3 Motivation!

4. 4 if - (x – M 1 ) T K 1 -1 (x – M 1 ) + (x – M 2 ) T K 2 -1 (x – M 2 ) > < C1C1 C2C2 T1T1 Optimum Decision Rules: 2-class Gaussian Quadratic Processing if ( M 1 – M 2 ) T K -1 x > < C1C1 C2C2 T2T2 Case 2: K 1 = K 2 = K Case 1: K 1 = K 2 Linear Processing Review 1

5. 5 if ( M 1 – M 2 ) T x > < C1C1 C2C2 T3T3 Case 3: K 1 = K 2 = K = s 2 I Optimum Decision Rules: 2-class Gaussian (cont) Linear Processing Review 2

6. 6 Q i (x) = (x – M j ) T K j -1 (x – M j ) } – 2 ln P(C j ) + ln | K i | M-Class General Gaussian MPE and MAP Select Class C j if Q j (x) is MINIMUM Select Class C j if L j (x) is MAXIMUM L j (x) = M j T K -1 x – ½ M j T K -1 M j + lnP(C j ) Case 2: K 1 = K 2 = … = K M = K Case 1: K 1 = K 2 Review 3

7. 7 Bayes decision rule is determined form a set of y i (x) defined by p(x|C k ) = 1 ( 2 ) N/2 KkKk ½ exp(- ½ (x – M k ) T K k -1 (x – M k ) ) C k : X ~ N( M k, K k ), P(C k ) M-Class General Gaussian: Bayes where Review 4

8. 8 (2 ) N/2 KjKj ½ C ij exp(- ½ (x – M j ) T K j -1 (x – M j )) P(C j ) y i (x) = j=1 M Taking the ln of the y i (x) for this case does not simplify to a linear or quadratic processor The structure of the optimum classifier uses a sum of exp( quadratic forms) and thus is a special form of nonlinear processing using quadratic forms. Review 5

9. 9 Gaussian assumptions Quadratic processing Linear and Reasons for studying linear, quadratic and other special forms of non linear processing If Gaussian we can find or learn a usable decision rule and the rule is optimum If non-Gaussian case we can find or learn a usable decision rule; however the rule is NOT necessarily optimum

10. 10 Linear functions f(x 1 ) = w 1 x 1 + w 2 One Variable f(x 1, x 2 ) = w 1 x 1 + w 2 x 2 + w 3 Two Variables f(x 1, x 2, x 3 ) = w 1 x 1 + w 2 x 2 + w 2 x 2 + w 3 Three Variables

11. 11 w 1 x 1 + w 2 = 0 w 1 x 1 + w 2 x 2 + w 3 = 0 w 1 x 1 + w 2 x 2 + w 3 x 3 + w 4 = 0 Constant Line Plane w 1 x 1 + w 2 x 2 + w 3 x 3 + w 4 x 4 + w 5 = 0 ? Answer = Hyperplane

12. 12 Hyperplanes w 1 x 1 + w 2 x 2 + … + w n x n + w n+1 = 0 x = [ x 1, x 2,…, x n ] T w 0 = [ w 1, w 2,…, w n ] T w 0 x + w n+1 = 0 Define An alternative representation of a Hyperplane is n-dimensional Hyperplane T

13. 13 Hyperplanes as boundaries for Regions R + = { x : } R - = { x: } Positive side of Hyperplane boundary Negative side of Hyperplane boundary w 0 x + w n+1 = 0 Hyperplane boundary

14. 14

15. 15 Definitions (1) Unit Normal u

16. 16 (2) Distance from a point y to the hyperplane (3) Distance from the origin to the hyperplane

17. 17 (4) Linear Discriminate Functions where Augmented Pattern Vector Weight vector

18. 18 Linear Decision Rule: 2-Class Case using single linear discriminant function No claim of optimality !!! for a vector x if given: d(x)=w 1 x 1 + w 2 x 2 + … + w n x n + w n+1

19. 19

20. 20 Linear Decision Rule: 2-Class Case using two linear discriminant function except on boundaries d 1 (x) = 0 and d 2 (x) = 0 where we decide randomly between C 1 and C 2 given two discriminant functions define decision rule by

21. 21 Decision regions (2-class case) using two linear discriminant functions and AND logic

22. 22 Decision regions (2-class case) using two linear discriminant functions(continued)

23. 23 Decision regions (2-class case) alternative formulation using two linear discriminant functions

24. 24 Decision regions (2-class case) using alternative form of two linear discriminant functions equivalent to

25. 25 Decision regions (3-class case) using two linear discriminant functions

26. 26 Decision regions (4-class case) using two linear discriminant functions

27. 27 Decision region R 1 (M-class case) using K linear discriminant functions

28. 28 Example: Piecewise linear boundaries Given the following discriminant functions

29. 29 If d 1 (x) > 0 AND d 2 (x) > 0 Define the following decision rule Show the decision regions in the two dimensional pattern space OR d 3 (x) > 0 AND d 4 (x) > 0 AND d 5 (x) > 0 AND d 6 (x) > 0 then decide x comes from class C 1, on the boundaries decide randomly, otherwise decide C 2 Example Continued

30. 30 Solution:

31. 31 Lecture 14 Summary 1. Reviewed structures of Optimal Classifier 2. Defined Linear functions, hyperplanes, boundaries, unit normals,various distances 3. Used Linear Discriminant functions for defining classifiers- Examples

32. 32 End of Lecture 14