Scribd
Upload
198 views

Pattern - Assigment1

This document provides details for an assignment in a pattern recognition course including 5 questions. Question 1 asks students to calculate the mean vector and sample covariance matrix for a set of samples and verify the calculations in Matlab. Question 2 asks students to show that the decision boundary for a minimum distance classifier is a hyperplane. Question 3 involves properties of multivariate normal distributions including eigenvalues, eigenvectors, and contour plots. Question 4 involves a linear transformation of a bivariate normal distribution. Question 5 involves calculating class means, covariances, and decision boundaries for a minimum distance and nearest neighbor classifier on a sample dataset.

Uploaded by

gizemyaren
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
198 views

Pattern - Assigment1

This document provides details for an assignment in a pattern recognition course including 5 questions. Question 1 asks students to calculate the mean vector and sample covariance matrix for a set of samples and verify the calculations in Matlab. Question 2 asks students to show that the decision boundary for a minimum distance classifier is a hyperplane. Question 3 involves properties of multivariate normal distributions including eigenvalues, eigenvectors, and contour plots. Question 4 involves a linear transformation of a bivariate normal distribution. Question 5 involves calculating class means, covariances, and decision boundaries for a minimum distance and nearest neighbor classifier on a sample dataset.

Uploaded by

gizemyaren
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

ENG 8801/9901 - Special Topics in Computer Engineering - Pattern Recognition!

Spring 2006

Assignment 1
Due May 18, 2006 at 7:00 PM

Question 1
10 marks

(a) Given the following set of samples, calculate the mean vector m and the sample covariance matrix Assignment 1 S. Formulas !

Charles Robertson May 8, 2005

(2,1), (3,2), (2,3), (1,2)

(b) Using Matlab, repeat the above calculations. Submit a print-out of your .m le. Question 2
10 marks

Show that the decision boundary dened by

Assignment 1 Formulas

|x m1 |2 = |x m2 |2

Charles Robertson (that is, the MED classier boundary) is given by the equation for a hyperplane, normal

to the line joining the two class means and passing through the midpoint of that May 8, straight 2005 line. Question 3
20 marks
2 |x m |2 = |x m 2| A1bivariate (i.e.

two variable) normal probability density function (p.d.f.) has mean (0,0)T and covariance matrix
2 1 1 2

(a) Find the eigenvalues and eigenvectors of the covariance matrix. (b) Calculate the equation of the 1-standard deviation contour of the p.d.f. (c) Sketch the 1-standard deviation contour (d) How are the eigenvalues and eigenvectors related to the shape of the 1-standard deviation contour?

May 8, 2005

ENG 8801/9901 - Special Topics in Computer Engineering - Pattern Recognition!


|xQuestion m1 |2 = |x4 m2 |2

Spring 2006

5 marks

Class 2 A has features which are normally distributed in the two-dimensional space. The 1 unit standard deviation contour has an area of 3 units. The data in class A are trans1 2 formed by the matrix:
3/5 4/5 4/5 3/5

What is the area of the unit standard deviation contour in the new space? Question 5
20 marks

You are given the following training sample patterns for two classes in feature space (x1, x2): Ci: (2,2) (4,0) (10,2) (8,4) (8,0) (4,4) Cj: (4,10) (8,18) (8,10) (6,6) (4,18) (6,22) (a) Calculate the class means and sample covariance matrices. Plot the samples. (b) On the plot, sketch the MED decision boundary using the class means as prototypes. (c) On the 1 same plot, sketch the MED decision boundary in the case where nearest neighbour classication is used. Note that the class prototypes will change! Clearly differentiate the two boundaries. (d) In which class will the new pattern (7,6) be classied where the prototypes are class means? Which class when the prototypes are decided by nearest neighbour classication?

You might also like