Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

UNIVERSITI TUN HUSSEIN ONN MALAYSIA

TEST (20%)
SEMESTER II
SESSION 2020/2021
COURSE NAME : INTELLIGENT CONTROL SYSTEM

COURSE CODE : BEH 41803/BEJ44103

4 BE4 BEJ PROGRAMME : 4 BEJ

DURATION : 1 HOUR AND 30 MINUTES

INSTRUCTION : 1. Do not look at or read the questions before you


are asked to do so.
2. Answer all questions
3. Cheating or other misconduct is not tolerated. In
doing so, you will automatically get zero mark for
this test.

Q1 /3

Q2 / 12

Q3 / 25

Total / 40
Q1. With an aid of illustration, explain why simple neural network (single neuron) can only be
used to classify linear sample set.
(3 marks)

Q2. The output equation for a single layer two inputs and one output Artificial Neural Networks
(ANN) is given below:
X1 X2 T
-1 -1 0
1 if 𝑊1 𝑋1 + 𝑊2 𝑋2 > 𝜃 -1 0 0
𝑌={
0 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒
0 0 0
0 1 1
1 0 1

Where W1 and W2 are signed integer weights, X1 and X2 are inputs, Y is output, T is target
and θ is threshold value. This network will be used to solve the sample in the table above.

a) Demonstrate all the sample data in a scatter plot of X1 versus X2 by labelling the ‘0’
output as ‘0’ and the ‘1’ output as ‘x’.
(2 marks)

b) By using analytical framework, identify the optimal weights, and threshold value that
can solve the problem with 100% of accuracy with verification steps after obtaining
the optimal parameters.
(10 marks)
Q3. As an engineer you are requested to design simple neural network (NN) model for clustering
six samples of three-bit pattern (x1, x2, x3) into cluster A and cluster B respectively as shown
below. For each pattern, grey colour box will be denoted as ‘0’ while white colour box as
‘1’. If cluster A will be labelled as ‘1’ and cluster B as ‘0’ during training:

a) Analyse an optimal NN decision boundary equation model after all six samples been
trained using Adaptive Linear Element (ADALINE) learning algorithm in its first epoch.
Use following table model with learning rate α = 0.5 in your training session.
(22 marks)
b) If new sample of pattern with value of x1= 1, x2=0, x3=1 was added to the cluster A.
With justification, analyse whether it still possible to train the sample set using
ADALINE network.
(3 marks)

X1 X2 X3 X1 X2 X3

Cluster A Cluster B

Iter X1 X2 X3 T S T-S W1 W2 W3 b
0 -1 -2 -1 1
1

You might also like