1 of 13

Neural Networks and Fuzzy Systems

Introduction

Rizoan Toufiq

Assistant Professor

Department of Computer Science & Engineering

Rajshahi University of Engineering & Technology

2 of 13

Introduction

  • Human and Computer:
    • Human beings are more intelligent than computers. Why is that said?
    • Calculating the sum of a few hundred eight and nine digit numbers
    • Does that make the computer more intelligent than us?
    • Consider: crosswords
      • Some of us are excellent
      • difficult for a computer
    • Computers are very bad at performing simple visual tasks.

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

2

3 of 13

Introduction

  • Human and Computer:
    • After 30 years of high-quality research, are not sufficient to allow them to make the claim that they have computer systems that are artificially intelligent in any general sense that we would recognize
    • The aim of artificial intelligence could be summed up as trying to make computers behave as they do in the movies.

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

3

4 of 13

Introduction

  • Human and Computer:
    • Why then can’t computers do the things that we do?
      • The nature of their design
      • Brain: slow, highly parallel; Computer: high-speed, serial machine
      • Counting is an essentially serial activity (Computer winner)
      • vision or speech recognition (Brain Winner)

​

  • [Note: Computers are designed to carry out one instruction after another, extremely rapidly, whereas our brains work with many more slower units. Whereas a computer can typically carry out a few million operations every second, the units in the brain respond about ten times per second.]

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

4

5 of 13

Introduction

  • Human~Computer:
    • Represent Complex information
    • Design Learning Algorithm
    • Find the technique of parallel processing

​

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

5

6 of 13

Introduction

  • The Structure Of The Brain:

​

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

6

7 of 13

Introduction

  • The Structure Of The Brain:

​

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

7

8 of 13

Introduction

  • The Structure Of The Brain:

​

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

8

9 of 13

Introduction

  • Learning In Machines
    • MENACE [Matchbox Educable Noughts and Crosses Engine]
    • Donald Michie (1960’s)
    • 288 Matchbox
      • Reinfocement learning take place
      • External Teachers are not required
      • Learn from examples

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

9

10 of 13

Introduction

  • The Differences
    • Brain Structure
    • Brain Performs in distributed processing manner
    • Fault Tolerance
    • Graceful degradation

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

10

11 of 13

Summary

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

11

  • Brain is parallel, distributed processing system.
  • Basic processing unit called the neuron.
  • Approximately 1010 neurons each connected to 104 others.
  • Operation of neuron: fires pulse down axon when sufficient input received from dendrites. Connections via chemical junctions called synapses.
  • Learning increases efficacy of synaptic junction.
  • Machines can learn through positive reinforcement.
  • Cerebral cortex shows local areas of specialized function.

12 of 13

Read Task

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

12

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

Chapter 1: Introduction

13 of 13

Query???

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

13