Perceptron Algorithm
Md. Zul Kawsar
130302005
Content
 Introduction
 Application
 Working Procedure
 Strength & Weakness
 Conclusion
 References
2
Introduction
 Perceptron is a simplest form of a neural network used for the classification of
patterns.
 The perceptron algorithm was invented in 1957 at the Cornell Aeronautical
Laboratory by Frank Rosenblatt.
 Frank Rosenblatt was a Psychologis.
Mark I Perceptron machine
The machine was connected to a camera that used
20×20 cadmium sulfide photocells to produce a 400-
pixel image
3
Introduction (Cont..)
 We Consider Linearly Separable Pattern as Example:
• Here S1 and S2 are two type of records .
4
Application
 In machine learning, the perceptron is an algorithm for supervised
classification of an input into one of several possible non-binary outputs.
 Perceptron can be defined as a single artificial neuron that computes its
weighted input with the help of the threshold activation function or step
function.
 The Perceptron is used for binary Classification.
 The Perceptron can only model linearly separable classes.
5
How it works
6
How it works (cont..)
7
How it works (cont..)
8
Strength and Weakness
 Strength
• Simple and computationally efficient.
• Guaranteed to learn a linearly separable problem.
 Weakness
• Only linear separations.
• Only converges for linearly separable data.
• Not really efficient with many features.
9
Conclusion
 Simple classification rule base on weight vector.
 Simple online learning algorithm guaranteed to converge if training set is
separable.
 With more than two class, perceptron uses one neuron to model a class against
the others.
10
References
 https://en.wikipedia.org/wiki/Perceptron
 Pattern Classification
Richard O.Dude
Peter E.Hart
11
Thank You
12

Perceptron algorithm

  • 1.
  • 2.
    Content  Introduction  Application Working Procedure  Strength & Weakness  Conclusion  References 2
  • 3.
    Introduction  Perceptron isa simplest form of a neural network used for the classification of patterns.  The perceptron algorithm was invented in 1957 at the Cornell Aeronautical Laboratory by Frank Rosenblatt.  Frank Rosenblatt was a Psychologis. Mark I Perceptron machine The machine was connected to a camera that used 20×20 cadmium sulfide photocells to produce a 400- pixel image 3
  • 4.
    Introduction (Cont..)  WeConsider Linearly Separable Pattern as Example: • Here S1 and S2 are two type of records . 4
  • 5.
    Application  In machinelearning, the perceptron is an algorithm for supervised classification of an input into one of several possible non-binary outputs.  Perceptron can be defined as a single artificial neuron that computes its weighted input with the help of the threshold activation function or step function.  The Perceptron is used for binary Classification.  The Perceptron can only model linearly separable classes. 5
  • 6.
  • 7.
    How it works(cont..) 7
  • 8.
    How it works(cont..) 8
  • 9.
    Strength and Weakness Strength • Simple and computationally efficient. • Guaranteed to learn a linearly separable problem.  Weakness • Only linear separations. • Only converges for linearly separable data. • Not really efficient with many features. 9
  • 10.
    Conclusion  Simple classificationrule base on weight vector.  Simple online learning algorithm guaranteed to converge if training set is separable.  With more than two class, perceptron uses one neuron to model a class against the others. 10
  • 11.
    References  https://en.wikipedia.org/wiki/Perceptron  PatternClassification Richard O.Dude Peter E.Hart 11
  • 12.