Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

United International University

Department of Computer Science and Engineering


CSI 415: Pattern Recognition Midterm, Summer 2015
Total Marks: 30 Time: 90 Minutes

Answer any 3 questions (3 × 10 = 30)


1. (a) Draw a diagram for Pattern Recognition System. Write short notes on, (1 + 6 = 7)
i. over-fitting
ii. missing feature
iii. evidence pooling
(b) In case of two classes case with zero-one loss function, show that minimizing risk is equivalent
to maximizing probability. (3)

2. A Credit Approval data set is given as follows that there are 3 attributes(Reference, Card Type, Avg
Expenditure) and 2 classes(positive, negative).

No. Reference Card Type Avg Expenditure Class


{refereed,non-refereed} {gold,silver} [0, 30k] {positive,negative}
1 refereed gold 9.4k positive
2 refereed gold 2k positive
3 refereed gold 24.9k negative
4 refereed silver 0k positive
5 refereed silver 15.6K negative
6 refereed gold 11.2k negative
7 refereed gold 5.5k positive
8 non-refereed silver 10.3k negative
9 refereed silver 6.2k positive
10 non-refereed silver 0k negative

Classify the following test data using,


(a) Bayesian Classifier (use only 5 bins to discretized continuous attribute) (5)
(b) Naive Bayesian Classifier (5)

Reference Card Type Avg Expenditure Class


refereed gold 9.4k ?

Given, Univarient normal distribution,


1 (Ai − µij )2
P (Ai |cj ) = p exp −
2πσij 2 2σij2

1
3. (a) State Celebrated Perceptron Algorithm. (5)
T
(b) Suppose, a perceptron is working on a 2 class problem with 3 attributes(X = [x1 , x2 , x3 ] ). At
t-th iteration, w0 = −0.5, w1 = 1, w2 = 1, w3 = 1. And in this iteration total 3 instances are
misclassified. They are
   
x1 −0.2
• x2  =  0.75 misclassified as class-2
x3 1.54
   
x1 0.15
• x2  =  0.4 misclassified as class-2
x3 0.034
   
x1 0.4
• x2  =  0.05 misclassified as class-1
x3 −1.45
What is the discriminator function in (t+1)-th iteration? (5)

4. (a) State Edit Distance Algorithm. What are Global and Local constraint of Edit Distance Algo-
rithm? (5 + 2 = 7)
(b) Explain Logarithmic Search. (3)

You might also like