Scribd
Upload
242 views

Artificial Neural Networks

Artificial neural networks are computer programs designed to learn like the human brain. They are composed of interconnected artificial neurons that can be trained to perform tasks. Neurons receive inputs, combine them using weights, and output a value based on an activation function. There are different types of neural networks based on their structure and learning method, including feedforward, feedback, supervised, unsupervised, and reinforcement networks. Neural networks are applied to problems like pattern recognition, forecasting, and control systems due to their ability to learn from examples without being explicitly programmed.

Uploaded by

punita singh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
242 views

Artificial Neural Networks

Artificial neural networks are computer programs designed to learn like the human brain. They are composed of interconnected artificial neurons that can be trained to perform tasks. Neurons receive inputs, combine them using weights, and output a value based on an activation function. There are different types of neural networks based on their structure and learning method, including feedforward, feedback, supervised, unsupervised, and reinforcement networks. Neural networks are applied to problems like pattern recognition, forecasting, and control systems due to their ability to learn from examples without being explicitly programmed.

Uploaded by

punita singh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 24

Artificial Neural Networks

Presented by- Punita singh


(1850211)
Ravi Ratnaik
(1850209)
Contents
 INTRODUCTION
 NEURONS
 ACTIVATION FUNCTION
 TYPES OF NEURAL NETWORKS
 LEARNING IN NEURAL NETWORKS
 APPLICATION OF NEURAL NETWORK
 ADVANTAGES OF NEURAL NETWORK
 DISADVANTAGES OF NEURAL NETWWORK
INTRODUCTION
 An artificial neural network can be seen as a computer
program that is designed to recognise patterns and learn
like the “human Brain”.
 ANNs like humans learn by examples and past experiences.
 It is composed of a large number of highly interconnected
processing elements(artificial neurons) working in unison
to solve specific problems.
 Learning in biological systems involves adjustments of
synaptic connections that exist between the neurons .This
is true for ANNs as well.
Continued..
 The first artificial neuron was formulated in 1943,by
neurophysiologist, McCulloch and logician Walter Pitts.

 Networks of the artificial neurons do not have even a


fraction of power of the Human Brain, but they can be
trained to perform useful functions.

 Neurons receive inputs from other sources, combine them


in some ways ,perform a generally non-linear operation on
the result and then output the final result.
Neurons
Description Neural Network Neuron
 Receives n inputs
 Multiplies each
input
 by its weight.

 Applies activation
function to the
sum of results.

 outputs results.
Activation Function
 The activation function falls in one of the following three
categories:-
a) Linear(or ramp)
b) Threshold
c) Sigmoid
 For linear units, the output activity is proportional to the
total weighted output.
 For threshold units, the output are set at one of the two
levels, depending on whether the total input is greater than
or less than some threshold value.
 For sigmoid units, the output varies continuously but not
linearly as the input changes.
Description Threshold activation Function
 The neuron will fire
output = 1, if
summation of XiWi is
positive.

 Output = 0 , for
unipolar

 Output = -1, for bipolar


if summation of XiWi is
negative.

 The sigmoid function


is = 1/1+e-net.
Types
According to topology-
 Feed forward ANN
 Feed back ANN

According to learning methods -


 Supervised
 Unsupervised
 Reinforcement
Continued..
Continued..
 The activity of the input units represents the raw information
that is fed into the network.

 the activity of each hidden unit is determined by the activities


of the input units and the weights on the connection between
the input and the hidden units.

 The behaviour of the output units depends on the activity of


the hidden units and the weights between the hidden and
output units.
Feed back ANN
Continued
 Feed back networks can have signals travelling in both the
directions by introducing loops in the network.

 Feed back networks are very powerful and can get extremely
complicated.

 Feed back networks are dynamic.

 Their state changes continuously until they reach an


equilibrium point.
Supervised Network
Continued..
 In this the network learns by comparing the network output
With the correct answer.

 The network receives feedback about the errors and weights


in different layers.

 And then adjusts its weights to minimize the error.

 This is also known as learning through a teacher or


reinforced learning.
Unsupervised Network
Continued..
 The network does not get any feedback about the errors.

 The network itself discovers the interesting categories or


Features in the input data.

 The unsupervised networks are expected to recognise the


input patterns , classify these on the basis of correlations

 And produce output signals corresponding to input


categories.
Back-Propagation Learning Algorithm
 Used for training feedforward neural networks

 Objective is to achieve the global minima

 It is supervised learning method

 We compare actual output of a network with desired output

 The network learns by propagating the instantaneous errors


back from each layer to its previous layer
Continued..
Training involves three stages:
 Feedforward of the input training pattern

 Calculation and back-propagation of the associated error

 Adjustment of weights

 Back-propagation algorithm is used for data modelling ,


forecasting , data and image compression
Applications
 Pattern recognition

 Banking

 Control systems and monitoring

 Financial Application

 Aerospace

 Forecasting – sales, market research and meteorology


Advantages
 Neural network can perform a task that a linear program
cannot do.

 When an element of a neural network fails , it can continue


its task without any problem due to their parallel nature.

 A neural network learns and need not to be reprogrammed.

 It can be implemented in any application.


Disadvantages
 The neural network needs training to operate.

 The architecture of neural networks is different from the

Architecture of microprocessors therefore needs to be


emulated.

 Requires high processing time for large neural networks.


REFERENCES
 Process control principles and applications by Surekha Bhanot

 www.slideshare.net
THANK YOU

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy