1 of 28

Neural Networks and Fuzzy Systems

The Basic Neuron

Rizoan Toufiq

Assistant Professor

Department of Computer Science & Engineering

Rajshahi University of Engineering & Technology

2 of 28

Modeling the single neuron

  • We can summarizes them as follows:
    • The output from a neuron is either on or off.
    • The output depends only on the inputs. A certain number must be on at any one time in order to make the neuron fire.

3 of 28

Modeling the single neuron

4 of 28

Modeling the single neuron

5 of 28

Modeling the single neuron

6 of 28

Modeling the single neuron

7 of 28

Modeling the single neuron

8 of 28

Learning in simple neuron

9 of 28

Learning in simple neuron

Our learning paradigm can be summarized as follows:

    • set the weights and thresholds randomly
    • present an input
    • calculate the actual output by taking the thresholded value of the weighted sum of the inputs
    • alter the weights to reinforce correct decisions and discourage incorrect decisions-i.e. reduce the error
    • present the next input etc �

10 of 28

The perceptron learning algorithm

11 of 28

The perceptron learning algorithm

12 of 28

The perceptron learning algorithm

13 of 28

The perceptron learning algorithm

14 of 28

The Perceptron: A Vectorial Perspective

15 of 28

The Perceptron: A Vectorial Perspective

16 of 28

The Perceptron: A Vectorial Perspective

17 of 28

The Perceptron Learning Rule: Proof

18 of 28

The Perceptron Learning Rule: Proof

19 of 28

The Perceptron Learning Rule: Proof

20 of 28

The Perceptron Learning Rule: Proof

21 of 28

The Perceptron Learning Rule: Proof

22 of 28

The Perceptron Learning Rule: Proof

23 of 28

The Perceptron Learning Rule: Proof

24 of 28

Limitations Of Perceptrons

25 of 28

Limitations Of Perceptrons

26 of 28

Summary

Rizoan Toufiq, Assistant Professor, Dept. of CSE, RUET

26

  • Perceptron-artificial neuron.
  • Takes weighted sum of inputs, outputs t1 if greater than threshold else outputs 0.
  • Hebbian learning (increasing effectiveness of active junctions) is predominant approach.
  • Learning corresponds to adjusting the values of the weights.
  • Feedforward supervised networks.
  • Can use +1, -1 instead of 0 , l values.
  • Can only solve problems that are linearly separable-therefore fails on XOR

27 of 28

Read Task

Rizoan Toufiq, Assistant Professor, Dept. of CSE, RUET

27

B1: Neural Computing - An Introduction - R Beale and T Jackson, Publisher: Adam Hilger, 1990 IOP Publishing Ltd.

Chapter 3: The Basic Neuron

28 of 28

Query???

Rizoan Toufiq, Assistant Professor, Dept. of CSE, RUET

28