Neural Networks and Fuzzy Systems - Engenharia … · Summary 1 - Introduction – Connectionist Intelligent Systems 2 - Artificial Neural Networks 3 - Fuzzy Logic and Fuzzy Systems

Soft Computing

Neural Networks and Fuzzy Systems

Prof. Dr.-Ing. Adolfo BauchspiessUniversidade de Brasília - Brazil

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

Summary

1 - Introduction – Connectionist Intelligent Systems2 - Artificial Neural Networks3 - Fuzzy Logic and Fuzzy Systems4 - Examples and Applications5 - Conclusions

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

5 - Conclusions

2

Part 1

Introduction – Connectionist Intelligent Systems

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 3

Some publications - Intelligent Systems

� Internacional Journals

� Neural Networks, IEEE Transaction son

� Fuzzy Systems, IEEE Transactions on

� Intelligent Systems Engineering

� Intelligent Systems, IEEE

�IntelligentTransportationSystems, IEEE Transactionson

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

�IntelligentTransportationSystems, IEEE Transactionson.....

� Conferences

.....

4

Epistemology –“Philosophy of Knowledge”

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 5

Heuristics

D Domain

Optimal Solution

D

S

H1H2

H3SolutionSpace

Heuristics give sub-optimal solutions.

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

Heuristics

Solution SpaceS

A heuristic rule leads from the domain space to the solution spaceD

S

H2

H3 Optimal SolutionSolution Space

H4H5H1

“Well-formed” Heuristics are close to the optimal solution.

6

Dislexy?

I cnduo't bvleiee taht I culod aulaclty uesdtannrd waht I wasrdnaieg. Unisg the icndeblire pweor of the hmuan mnid, aocdcrnig to rseecrahat Cmabrigde Uinervtisy, it dseno'tmttaer

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

aocdcrnig to rseecrahat Cmabrigde Uinervtisy, it dseno'tmttaerin waht oderr the lterets in a wrod are, the olny irpoamtnt tihngis taht the frsit and lsat ltteer be in the rhgit pclae. The rset canbe a taotl mses and you can sitll raed it whoutit a pboerlm. Tihs isbucseae the huamn mnid deos not raed ervey ltteer by istlef, butthe wrod as a wlohe. Aaznmig, huh? Yaeh and I awlyas tghhuotslelinpg was ipmorantt! See if yuor fdreins can raed tihs too.

7

Giant x 3D Ilusion?

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 8

Waves?

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 9

Simpathic?

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 10

Antipathic?

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 11

Introduction - Connectionist Intelligent Systems

Artificial IntelligenceScience field that studies paradigmsthat aims to explain howintelligent behaviourcan emerge

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

can emergefrom artificial implementations, in computers.

Intelligence: learning, adaptation, comprehension

12

IA Paradigms

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

Intelligence : learning, adaptation, comprehension

13

Connectionist Paradigm

Considers to be virtually impossible to transform in algorithms -i.e., to reduce to a sequence of logical and arithmetic stepsmany tasks that the human mind performs with ease and speed, for example:

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

· Face Recognition,· Comprehend and translate natural languages,· Memory evocation by association,· Games...

for example:

14

Connectionist Paradigm

The computational process have to mimicthe brain capacity of self-organization → learn!

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 15

Symbolist versus ConectionistParadigm

-Perception

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

The Kanizsa square, 1976

16

“Local Coherency –Global Paradox"

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 17

Sandro delPrete. Enigmas Visuais. Rio de Janeiro, 2004, p. 45

M.C.Escher

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

“Positive Truth X Negative Truth”

18

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

“Up stairsX

Down stairs”

19

Paradigma Simbolista versus Conexionista

- J.S. Bach “Coerência Local - Paradoxo Global”

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

Pseudo-rising scale

played on a vibraphone

Shepard's scale

20

Synapses Formation

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic

0-2 years 2 years to puberty Adult

21

Olfative Information Processing

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 22

Auditive Information Processing

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 23

Epilepsy Pacient –without left brain hemisphere since 12 years age

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 24

A neuron over a Chip.

Laboratório de Automação e Robótica - A. Bauchspiess– Soft Computing - Neural Networks and Fuzzy Logic 25