Scribd
Upload
95 views

Artificial Neural Networks

Artificial neural networks are computational systems inspired by biological neural networks in the brain. They consist of artificial neurons that receive inputs, process them, and transmit outputs. The perceptron is an artificial neuron that has input connections that receive weighted inputs, a processor that sums the inputs and applies an activation function, and an output connection that transmits the output. Neural networks learn by adjusting the weights between connections through supervised, unsupervised, or reinforced learning. The perceptron learning algorithm uses feed-forward and backpropagation methods to iteratively adjust weights to reduce errors between actual and expected outputs until the network is adequately trained.

Uploaded by

Pavan Sai
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
95 views

Artificial Neural Networks

Artificial neural networks are computational systems inspired by biological neural networks in the brain. They consist of artificial neurons that receive inputs, process them, and transmit outputs. The perceptron is an artificial neuron that has input connections that receive weighted inputs, a processor that sums the inputs and applies an activation function, and an output connection that transmits the output. Neural networks learn by adjusting the weights between connections through supervised, unsupervised, or reinforced learning. The perceptron learning algorithm uses feed-forward and backpropagation methods to iteratively adjust weights to reduce errors between actual and expected outputs until the network is adequately trained.

Uploaded by

Pavan Sai
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/ 8

ARTIFICIAL NEURAL NETWORKS

Introduction

BIOLOGICAL
NEURON

 Neural Networks : A machine that is designed to model the way in which the brain performs a
particular task or function or function of interest.

 Artificial Neurons are crude approximations of the neurons found in Human Brains.

 From a practical point of view, an ANN is just a parallel computational system consisting of many simple
processing elements connected together in simple processing elements connected together in a
specific way in order to perform a particular task.
The Neuron - A Biological Information Processor
• Dendrites - the receivers
• Soma - neuron cell body (sums input signals)
• Axon - the transmitter
• Synapse - point of transmission
• Neuron activates after a certain threshold is met
Learning occurs via electro-chemical changes in effectiveness of synaptic junction.
An Artificial Neuron - The Perceptron
• Simulated on hardware or by software
• Input connections - The Receivers
• Node (Or) Unit - Simulates neuron body
• Output connection - The Transmitter
• Activation function employs a threshold or bias
• Connection weights act as synaptic junctions
Learning occurs via changes in value of the connection weights.
Basic REPRESENTATION of a perceptron:

The function of perceptron is classified into 3 parts:


1) Receiver - It receives the weighted outputs of the previous layer in the network
2) Processor - It sums up the inputs received from the previous layer and applies the activation function
to scale the net sum between 0 and 1.
3) Transmitter – It transmits the output from the processor to the next level by multiplying the weight
associated with that link.
 Activation Function:-
In artificial neural networks, the activation function of a node defines the output of that
node, or "neuron," given an input or set of inputs. This output is then used as input for the
next node and so on until a desired solution to the original problem is found.
It maps the resulting values into the desired range such as between 0 to 1 or -1 to 1 etc.
(depending upon the choice of activation function).
 Necessity:-
An activation function is biologically inspired from the activities in our brain where
different neurons fire or get activated by different stimuli. This firing of a neuron can be
represented by ‘1’ for firing and ‘0’ for not firing.
The activation function gives us an indication of to what extent the particular neuron is
activated to the given input which is quite necessary to understand about the role of a
neuron in giving that particular output.
A typical Neural Network looks as follow :
 As explained, the output of neural network depends on the weights associated with the links between
nodes in various layers.

 To make a neural network function properly, it is vital to train it. Learning is of 3 types:

(1)Supervised learning- Requires teacher (i.e., Requires input vectors and target vectors). Similar to our
learning experience in childhood.
(2)Unsupervised Learning- It is all about discovering patterns in the data available without any prior
knowledge about it and no help from the teacher. Similar to our learning experience in adulthood.
(3)Reinforced Learning- A ‘teacher’ though available, does not present the expected answer but only
indicates if the computed output is correct or incorrect. It is a kind of trial and error algorithm. This
has its applications in simulating board games like chess.

NOTE: The difference between Supervised


learning and Reinforced learning is that in
the later ones, the learner is not told what
actions to take in a particular environment
but instead, must discover which actions
yield the most reward by trying them.
Perceptron Learning Algorithm:
Two mechanisms are necessary to form a trained neural network:
1) Feed-Forward
2) Back Propagation.
• Feed Forward is the mechanism where the inputs are passed into the neural network, processed (i.e.,
corresponding activation is found) and then forwarded to the next layer. This is first step in learning.
• Now the output is compared to the expected output and undergoes back propagation( Explained below) to
reduce the error.
• Back Propagation is the algorithm used to change the weights of the links in the network according to the input
and corresponding output. It is the error ( Difference between actual output and expected output) that is back
propagated through the network through which the weights are altered accordingly (i.e., increased or decreased
based upon the existing weights)
The following process should be employed in the network until a well trained network is formed
(i.e., the error is below an acceptable level) :
1. Initialize weights randomly.
2. Present a pattern and target output
3. Compute output
4. Update weights
The steps 2-4 are repeated until the required condition is met.

You might also like