You are on page 1of 1

BIRLA INSTITUTE OF TECHNOLOGY & SCIENCE, PILANI

Pilani Campus
Semester I: 2021-2022

MEL G622: Introduction to Artificial Neural Networks

Lab 4: Support vector classification and Multi-layer Perceptron

Done (From previous lab): Note: Use part [A] which is done in previous lab to
perform part [B] .

[A] Design a single layer perceptrons to classify two different sets of data patterns, based
on the measurements of ‘x1’ and ‘x2’. Assume that each of the data pattern contains 50
data points. Create your own two sets of data patterns randomly and then design your
perceptron both using the MATLAB command and toolbox. Show the complete analyses
and make your observations. Use the figures to illustrate your points.

To be done (in this lab):

[B] Now, repeat part (A), with using the support vector classification with three support
vectors. Make your observations and compare your findings from (A)?

[C] Use the attached “data1.xls”(access this file via http://tiny.cc/DataSet1) to classify the
dataset into three different classes (e.g., C1, C2, and C3). Use feedforward network
structure, with back propagation algorithm, for the classification. Assume supervised
learning and you can select the target values (i.e., first 50 samples correspond to C1, the
next 50 samples correspond to C2 and the remaining samples correspond to C3).

Draw the network structure with all the necessary weights and bias values after the
training. You will have to use the thresholding operation to the network output. Make
your observations. Use the MATLAB’s neural network GUI to simulate the feedforward
network. Compare your observations when you vary the number of neurons in the hidden
layer.

You might also like