Soft computing

1. SOFT COMPUTING PRESENTED BY: GANESH PAUL TT – IT(02)

2. What is Soft Computing? Soft computing is an emerging approach to computing which parallel the remarkable ability of the human mind to reason and learn in a environment of uncertainty and imprecision. Some of it’s principle components includes: Neural Network(NN) Fuzzy Logic(FL) Genetic Algorithm(GA) These methodologies form the core of soft computing.

3. GOALS OF SOFT COMPUTING The main goal of soft computing is to develop intelligent machines to provide solutions to real world problems, which are not modeled, or too difficult to model mathematically. It’s aim is to exploit the tolerance for Approximation, Uncertainty, Imprecision, and Partial Truth in order to achieve close resemblance with human like decision making.

4. SOFT COMPUTING - DEVELOPMENT HISTORY Soft = Evolutionary + Neural + Fuzzy Computing Computing Network Logic Zadeh Rechenberg McCulloch Zadeh 1981 1960 1943 1965 Evolutionary = Genetic + Evolution + Evolutionary + Genetic Computing Programming Strategies programming Algorithms Rechenberg Koza Rechenberg Fogel Holland 1960 1992 1965 1962 1970

5. NEURAL NETWORKS An NN, in general, is a highly interconnected network of a large number of processing elements called neurons in an architecture inspired by the brain. NN Characteristics are:- Mapping Capabilities / Pattern Association Generalisation Robustness Fault Tolerance Parallel and High speed information processing

6. 6 Neuron Biological neuron Model of a neuron

7. ANN ARCHITECTURES Input Layer Output Layer 1.Single Layer Feedforward Network Input Layer Hidden Layer Output Layer 2.Multilayer Feedforward Network Input Layer Hidden Layer Output Layer 3.Recurrent Networks Xi - Input Neuron Yi - Hidden /Output Neuron Zi - Output Neuron i = 1,2,3,4….. X1 X2 X3 y1 y2 y3 X1 X2 X3 y1 y2 z1 z2 z3 X1 X2 X3 y1 y2 z1 z2 z3

8. LEARNING METHODS OF ANN NN Learning algorithms S Supervised Learning Unsupervised Learning Reinforced Learning Error Correction Stochastic Hebbian Competitive Least Mean Square Backpropagation

9. FUZZY LOGIC Fuzzy set theory proposed in 1965 by A. Zadeh is a generalization of classical set theory . In classical set theory, an element either belong to or does not belong to a set and hence, such set are termed as crisp set. But in fuzzy set, many degrees of membership (between o/1) are allowed

10. FUZZY VERSES CRISP FUZZY CRISP IS R AM HONEST ? IS WATER COLORLESS ? FUZZY CRISP Extremely Honest(1) Very Honest(0.8) Honest at Times(0.4) Extremely Dishonest(0) YES!(1) NO!(0)

11. OPERTIONS CRISP FUZZY 1.Union 2.Intersection 3.Complement 4.Difference 1.Union 2.Intersection 3.Complement 4.Equality 5.Difference 6.Disjunctive Sum

12. PROPERTIES CRISP FUZZY Commutativity Associativity Distributivity Idempotence Identity Law Of Absorption Transitivity Involution De Morgan’s Law Law Of the Excluded Middle Law Of Contradiction Commutativity Associativity Distributivity Idempotence Identity Law Of Absorption Transitivity Involution De Morgan’s Law

13. GENETIC ALGORITHM Genetic Algorithms initiated and developed in the early 1970’s by John Holland are unorthodox search and optimization algorithms , which mimic some of the process of natural evolution. Gas perform directed random search through a given set of alternative with the aim of finding the best alternative with respect tp the given criteria of goodness. These criteria are required to be expressed in terms of an object function which is usually referred to as a fitness function.

14. BIOLOGICAL BACKGROUND All living organism consist of cell. In each cell, there is a set of chromosomes which are strings of DNA and serves as a model of the organism. A chromosomes consist of genes of blocks of DNA. Each gene encodes a particular pattern. Basically, it can be said that each gene encodes a traits. Fig. Genome consisting Of chromosomes. A T G C T A G C A G T A C

15. ENCODING There are many ways of representing individual genes. Binary Encoding Octal Encoding Hexadecimal Encoding Permutation Encoding Value Encoding Tree Encoding

16. BENEFITS OF GENETIC ALGORITHM Easy to understand. We always get an answer and the answer gets better with time. Good for noisy environment. Flexible in forming building blocks for hybrid application. Has substantial history and range of use. Supports multi-objective optimization. Modular, separate from application.

17. APPLICATION OF SOFT COMPUTING Consumer appliance like AC, Refrigerators, Heaters, Washing machine. Robotics like Emotional Pet robots. Food preparation appliances like Rice cookers and Microwave. Game playing like Poker, checker etc.

18. FUTURE SCOPE Soft Computing can be extended to include bio- informatics aspects. Fuzzy system can be applied to the construction of more advanced intelligent industrial systems. Soft computing is very effective when it’s applied to real world problems that are not able to solved by traditional hard computing. Soft computing enables industrial to be innovative due to the characteristics of soft computing: tractability, low cost and high machine intelligent quotient.

19. REFERENCES Neural Networks, Fuzzy Logic, and Genetic Algorithms Synthesis and Application by S. Rajasekaran and G.A. Vijayalakshmi Patel L. A. Zadeh, “Fuzzy logic, neural networks and soft computing,” in Proc. IEEE Int. Workshop Neuro Fuzzy Control, Muroran, Japan, 1993. T. Nitta, “Application of neural networks to home appliances,” in Proc. IEEE Int. Joint Conf. Neural Networks, Nagoya, Japan, 1993. P.J. Werbos, “Neuro-control and elastic fuzzy logic: Capabilities, concepts and application,” IEEE Trans. Ind. Electron., Vol. 40. 1993. Y. Dote and R.G. Hoft, Intelligent Control-Power Electronics Systems. Oxford, U.K.: Oxford Univ. Press, 1998. L. A. Zadeh, “From computing with numbers to computing with words-From manipulation of measurements to manipulation of perception,” IEEE Trans. Circuit Syst., Vol. 45, Jan 1999.

20. Any Questions