This document provides an introduction to neural networks, including: - An overview of neurons and how they are connected in a neural network. - The main types of neural networks including feedforward, feedback, single layer, and multilayer networks. - How neural networks learn through supervised, unsupervised, and reinforcement learning methods. - Common applications of neural networks like pattern recognition, control systems, and forecasting. - The advantages of neural networks in tasks linear programs cannot do and their ability to learn without reprogramming. Disadvantages include the need for training and high processing requirements for large networks.

1. I ~ - . ' ~ I
Guifu.~ ~ ~
Chapala Mah~rana·-'
.,,,,,.. i'
' ' '
J
· SubmitteJ '
. .
I
N .

2. Contents
. Introduction to Neural Network
.Neurons
.Activation Function
.Types of Neural Network
.Learning In Neural Networks
. Application of Neural Network
.Advantages of Neural Network
.Disadvantages Of Neural
Network

3. Neural Networks
• A method of computing, based on the
interaction of multiple connected
processing elements .
• A powerful technique to solve many real
world problems .
• The ability to learn from experience in
order to improve their performance.
• Ability to deal with incomplete
info tion

4. Basics Of Neural Network
• Biological approach to AI
• Developed in 1943
• Comprised of one or more
layers of neurons
• Several types, we'll focus on
feed-forward and feedback
networks

5. Nucleu1
Cell llody
Mylllln
11--Axon
lynap(lc teminll
1
Neurons
Threshold luoclion
Neural Network
Neuron

6. Neural Network Neurons.
• '.· ' • ' • • • • + ' • : • - • : • : ' ••
• • • • • • ._• + • ~ ••
. . . . . . . . . . . '

7. Activation Functions
•
11------
•
...............•
•2 •I 2
------'
.·I

8. ,,:::__
'.l'yp_
es_ of Neural Networ
- -
Neural Network types can be classified based on following
attributes:
•CoM•otJ..on
Static (feedforward)
- Dynamic (feedback)
• 'l'opolOff
Single layer
Multilayer
Recurrent
. ,......,_.. ........
Supe rViiiO:"d
tJni,upervised
·--•- Reinforcement

9. Cl assifi ca ti on Based On Connect i o n Types
•S L'."i : i .· (F--- l ! . ·r ·.•: c1 r ·i )
Class i fi ~a : i o n Based On Topo logy
ftlctef . I ~
I ,I, ii 'I I
,-.· I, , ,·, '' l
I
j ~ , I , I'
1l i .!1111 .
'- , ~ , Io..' 11 l l

10. .II
Classification Based On Learning Method
• Supervised
• Unsupervised
• Reinforcement
Supervised learning
• Each training pattern: input+ desired
output
• At each presentation: adapt weights
• After many epochs convergence to a
local minimum
4---
iI .

11. Unsupervised Learning
• No help from the outside
• No training data, no information
available on the desired output
• Learning by doing
• Used to pick out structure in the
input:
• Clustering
• Reduction of dimensionality ➔
compression
• Example: Kohonen's - Learning Law

12. Reinforcement learning
• Teacher: training data
• The teacher scores the performance of the
training examples
• Use performance score to shuffle weights
'randomly'
• Relatively slow learning due to
'randomness'
••
.,

13. Neura1 Network A 1ications
• Pattern recognition
• Investment analysis
• Control systems &
monitoring
• Mobile computing
• Marketing and financial
applications
• Forecasting - sales , market
research , meteorology

14. antages :
•A neural network can perform
linear program can not .
tasks that a
•When an element of the neural network
fails , it can continue without any problem
by their parallel nature.
" •A neural network learns and does not need
to be reprogrammed .
•It can be implemented in any application.
--·•· -
•It can
I ,,

15. Disadvantages:
•The neural network needs training
to operate .
•The architecture of a neural
network is different from the
architecture of microprocessors
therefore needs to be emulated.
•Requires high processing
-ff~'- 1arge neural networks.
,I
time for

16. Conclusions
• Neural networks provide ability to
provide more human- like AI
• Takes rough approximation and
hard- coded reactions out of AI
design (i.e . Rules and FSMs)
• Still require a lot of fine - tuning
during development

17. References
• Neural Networks , Fuzzy Logic, and Genetic
Algorithm ( synthesis and Application)
S . Rajasekaran , G. A. Vijayalakshmi Pai, PHI
• Neuro Fuzzy and Soft Computing, J. S. R.
JANG,C.T. Sun , E. Mitzutani, PHI
• Neu r al Netware , a tutorial on neural networks
• Sweetser , Penny . "Strategic Decision- Making with
Neural Networks and Influence Maps", AI Game
Pro rammin Wisdom 2 , Section 7.7 (439 - 46)
• Russell , Stuart and Norvig, Peter . Artificial
Intelli ence : A Modern A roach, Section 20 . 5
(736 - 48)