Sei sulla pagina 1di 6

COMP 304 – ARTIFICIAL

INTELLIGENCE
ASSIGNMENT 2: BACKPROPAGATION NETWORKS

L.L Vuso 200308237

Table of Contents
· Definitions and Acronyms
· Problem 1( Male/Female Classification)
· Problem 2 (Rabbits)
Definitions and Acronyms

· Neural Networks( NN ) : computational model that is inspired by the structure


and/or functional aspects of biological neural networks. It consists of an interconnected
group of artificial neurons and processes information using a connectionist approach to
computation.
·

Problem 1:Male / Female classification

Cross Validation and Test Data

NN 1 NN 2 NN
3
% of training data for cross validation 20 20 20
% of training data for Test 20 20 20

Multilayer

Hidden Layers 1 2 0

Hidden Layer Number 1

Processing Elements 9 9 -
Transfer TahnAxon TahnAxon -
Learning Rule Momentum Momentum -
Step Size 1.000 1.500 -
Momentum 0.700 0.700 -

Hidden Layer Number 2

Processing Elements - 4 -
Transfer - TahnAxon -
Learning Rule - Momentum -
Step Size - 0.100 -
Momentum - 0.700 -

Output Layer
Transfer TanhAxon TahnAxon TahnAxon
Learning Rule Momentum Momentum Momentum
Step Size 1.000 0.010 1.000
Momentum 0.700 0.700 0.700

Supervised Learning Control

Maximum Epochs 1000 1000 1000


Weight Update Batch Batch Batch

Output Table 1

Height Weight Gender


160 60 F
160 70 F
160 80 F
165 60 F
165 70 F
165 80 M
170 60 F
170 70 M
170 80 M

Problem 2 :Rabbits

Cross Validation and Test Data


NN 1 NN 2 NN 3
% of training data for cross validation 20 20 20
% of training data for Test 20 20 20

Multilayer

Hidden Layers 4 2 0

Hidden Layer Number 1

Processing Elements 4 2 -
Transfer TahnAxon TahnAxon -
Learning Rule Momentum Momentum -
Step Size 1.000 1.000 -
Momentum 0.700 0.700 -

Hidden Layer Number 2


Processing Elements 4 2 -
Transfer TahnAxon TahnAxon -
Learning Rule Momentum Momentum -
Step Size 0.100 0.100 -
Momentum 0.700 0.700 -

Hidden Layer Number 3

Processing Elements 4 - -
Transfer TahnAxon - -
Learning Rule Momentum - -
Step Size 0.010 - -
Momentum 0.700 - -

Hidden Layer Number 4

Processing Elements 4 - -
Transfer TahnAxon - -
Learning Rule Momentum - -
Step Size 0.001 - -
Momentum 0.700 - -

Output Layer

Transfer TanhAxon TahnAxon TahnAxon


Learning Rule Momentum Momentum Momentum
Step Size 0.001 0.010 1.000
Momentum 0.700 0.700 0.700

Supervised Learning Control

Maximum Epochs 1000 1000 1000


Weight Update Batch Batch Batch

Output Table 2
Age Weight
400 204.6776
450 218.6220
500 225.8452
550 203.4586
600 214.8147
850 221.7161

Potrebbero piacerti anche