. } | ap @ less than PHOTOCOPY price VMNsUBURCT CODE) 410282(0) —_ 6 gyllabut 1) = 2015 Sfattet = srt \ rmyaterd S80 \ ter sat . DECODE © greatt Sy SOFT COMPUTING AND OPTIMIZATION ALGORITHMS Iresh A. Dhotre ME. (Information Technology) &x-Faculty, Sinhgad College of Engineering Pune Written by Popular Authors of Text Books of Technical Publications Covers Entire Syllabus Question - Answer Format Exact Answers & Solutions | | @ Less than PHOTOCOPY Price |SYLLABUS OFT COMPUTING AD OVtiaization ALGORITHMS (41075715),, Unit 1s Introduction Introduction, HAL COMING YH ane iind nese YR Uh rye rechniques, and applications A UM anyriting base ele f yh oe ¢ Fuzzy logic, neural netwer emduianmry op 4 Setters Sersent networks, application wpe A natal nermabs, Vary bage. qarare weerien ned hybrid systems. (Chapter - 4) Unit Il: Fuzzy Sets and Logic basic concepts of fuzzy Ingic, Fuzzy wets and operations, Properties of fuzzy wets, Buzzy and conversion. Membership functions, tn: Fuzzy implications and Fuzzy algentir (Chapter - 2) Unit IIT: Fuzzy Systems Puzzy Controller, Fuzzy rule base and appromimae rence tables in fuzzy logic, fuzzy propenitions formation of +t compound rules, aggregation of fuzzy rules, fuzzy ren fuzzy expert systems, (Chapter - 3) Unit IV : Evolutionary Computing Basic Evolutionary Processes, EV: A Simple Evolutionary Spsem, Systems as Problem Solvers, A Historical Perspect: Aigorithms - Evolutionary Programming. Evolution Stretegies, A U. Simple EAs- A Common Framework, Population Size. (Chapter - 4) Unit V : Genetic Algorithm Basic concepts, working principle, procedures of GA. flow che of GA. Genetic representations, (encoding) Initialization and selection, Genetic operators, Mutation. Generational Cycle, Traditional algorithm vs genetic algorithm, simpie GA. general genetic algorithm, schema theorem, Classification of genetic algorithm, Hollan: classifier systems, genetic programming. applications of genetic aigorith Convergence of GA. Applications and advances in GA, Differences and similarities between GA and other traditional method, applications. (Chapter - 5) Unit VI: Swarm Intelligence Swarm intelligence, Particle Swarm Optimization (PSO) Algorithm- Formulations. Psetido-code, parameters, premature convergence, topology. biases, Real valued and binary PSO, Ant colony optimization (ACO) Formulations, Pseudo-code. Applications of PSO and ACO. (Chapter - 6) ee asi) fuzzy inference system.TABLE OF CONTENTS “ Chapter-1 Introduction (1-4) to(4. a ) 1.1 Introduction of Soft Computing . . 1.2. Constitutes of Soft Computing. .. der 1.3. Applications of Sof Computing .. Leg 1.4 Basic Tools of Soft Computing . Le In Introduction : Neural Networks . Chapter-2 Fuzzy Sets and Logic (2 - 1) to (2 - 32) 2.1. Basic Concepts of Fuzzy Logic .......s.0ee0eeeeeeeeees 2.2. Fuzzy Sets and Crisp Sets a 2.3. Fuzzy Set Theory and Operations . 2.4 Fuzzy and Crisp Relations, Fuzzy to Crisp Conversion........+.+0eeeeeeeereeeeeee 2.5 Interference in Fuzzy Logic 2.6 Fuzzy if-then Rules 7 Fuzzyfications and Defuzzifications ..............6++555 ) Chapter-3 Fuzzy Systems 3.1 Fuzzy Controller. .........ssssee eee ee eee een e eee e es 3,2 Fuzzy Rule Base and Approximate Reasoning .......-+++++ 3.3 Pecom cation of Compound Rules and Aggregation of 3-1 11 (nnw 3.4 Fuzzy Inference System 3.8 Furzy Expert Systems Chapter -4 — Evolutionary Computing (4-1) to (4-5) 4.1 Basic Evolutionary Proc 4.2) Evolutionary Systems Problem Solv 4.3) Canonical Evolutionary Algorithm Chapter-5 Genetic Algorithm (5 - 1) to (5 - 39) 5.1 Basic Concepts and Working Principle ................... 5-1 5.2. Genetic Operators. $.3. Traditional Algorithm vs Genetic Algorithm. 5.4 Simple GA and General Genetic Algorithm 5.5 Schema Theorem 5.6 Classification of Genetic Algorithm 5.7 Holland Classifier Systems 5.8 Genetic Programming ........... 0.00. ce cece eee eee 5.9 Applications of Genetic Algorithm . 5.10 Convergence of GA... Chapter-6 Swarm Intelligence (6 - 1) to (6 - 17) 6.1 Introduction of Swarm Intelligence ........ 6660s eee eee e ee 6.2 Particle Swarm Optimization (PSO) Algorithm. 6.3 Ant Colony Optimization a1: Introduction of Soft Computing Q.1 Describe what is soft computing. Explain its goals. Ans.: ¢ Soft computing is likely to play an especially importar, role in science and engineering, but eventually its influence may extend much farther. Building human-centered systems is ar imperative task for scientists and engineers in the new millennium, ¢ Soft computing (SC ) is a concept that was introduced by Zadet (1992 ), the discoverer of fuzzy logic. © Definition : "Soft computing is an emerging approach to computing which parallel the remarkable ability of human mind to reason and learn in a environment of uncertainty and imprecision". © The idea behind soft computing is to model cognitive behavior of human mind. Soft computing is foundation of conceptual intelligence in machines. Unlike hard computing, soft computing is tolerant of imprecision, uncertainty, partial truth, and approximation. ¢ Hard computing requires a precisely stated analytical model ani often a lot of computation time. Many analytical models are valid for ideal cases. © Soft computing is a collection of methodologies that aim to exploit the tolerance for imprecision and uncertainty to achieve tractability, robustness, and low solution cost. a-1Soft Computing & Optimization Algorithms 1-2 Introduction ¢ Soft computing is not precisely defined. It consists of distinct concepts and techniques which aim to overcome the difficulties encountered in real world problems. These problems result from the fact that our world seems to be imprecise, uncertain and difficult to categorize. The basic methods included in cognitive computing are fuzzy logic, neural networks and genetic algorithms, the methods which do not derive from classical theories. © Soft computing combines knowledge, techniques, and methodologies from the sources above to create intelligent systems. © The guiding principle of soft computing is : Exploit the tolerance for imprecision, uncertainty, partial truth, and approximation to achieve tractability, robustness and low solution cost. © Soft computing does not perform much symbolic manipulation, so user can view it as a new discipline that complements conventional artificial intelligence approaches and vice versa. «Hard computing is also called as conventional computing. It requires a precisely stated analytical model and often a lot of computation time. Many analytical models are valid for ideal cases. Real world problems exist in a non-ideal environment. Unique property of soft computing © Learning from experimental data. © Soft computing techniques derive their power of generalization from approximating or interpolating to produce outputs from previously unseen inputs by using outputs from previous learned inputs. © Generalization is usually done in a high-dimensional space. Goal of soft computing e It is a new multidisciplinary field, to construct a new generation of Artificial Intelligence, known as Computational Intelligence. than PHOTOCOPY PriceSoft Computing & Optimization Algorithms 1-3 Introduc Un The main goal is to develop intelligent machines to provig, solutions to real world problems, which are not modeled or ¢, difficult to model mathematically. elts aim is to develop the tolerance for Approximation, Uncertainty, Imprecision, and Partial Truth in order to achiey, close resemblance with human like decision making. Q.2 List and characterize the constituents of soft computing, Ans, : 1. One of the characteristics of soft computing methods is that they are typically used in problems where mathematical models are not available or are intractable or too cumbersome to be viable. 2. Another characteristic is that uncertainty inherent in many situations under study is taken into account rather than ignored, 3. Soft computing often provides a good solution as opposed to an optimal solution. 4. Soft computing uses human concept like if-then rules, cases and conventional knowledge representations. 5. Biologically inspired computing models (NN). 6. New optimization techniques are applied by soft computing. 7. Model free learning : Models are constructed based on the target system only. 8. New application domains : Mostly computation intensive like adaptive signal processing, adaptive control, nonlinear system identification etc. 9. Fault tolerance : Deletion of a neuron or a rule does not destroy the system. The system performs with lesser quality. 10.Goal driven characteristics : Only the goal is important and not the path, @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms 1-4 Introduction Q.3 Compare hard computing and soft computing. Ans. : Hard computing Soft computing It uses binary logic. It is based on numerical analysis. It uses fuzzy logic. It is based on genetic algorithms. Crisp system is used in had computing. Neurocomputing is used in soft computing. It takes help of differential equations. It takes help of probabilistic reasoning. Approximation theory. Evidential reasoning. Functional analysis. Machine learning. Funcvona ‘Approximate, spproxnaton reasoning ard randomized Q.4 List strength and weakness of soft computing. Ans. : Strength : Low cost and less time consuming solutions are provided by 1. soft computing. . Soft computing techniques can tolerate imprecision, uncertainty and partial truth to produce High Machine Intelligent Quotient. . It provides good solutions. . Tt uses human concept. @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms | 1-5 Introduction Weakness : 1. It requires extensive computation. 2. Soft computing does not perform much symbolic manipulation, 3. Soft computing is not precisely defined. .2 : Constitutes of Soft Computing Q.5 Describe fuzzy approach of soft computing. Ans.: © Fuzzy Logic (FL) is a multivalued logic, that allows intermediate values to be defined between conventional evaluations like true/false, yes/no, high/low, etc. Fuzzy Logic is based on fuzzy set theory and provides methods for modeling and reasoning under uncertainty, a characteristic present in many problems, which makes FL a valuable approach. © It allows data to be represented in intuitive linguistic categories instead of using precise (crisp) numbers which might not be known, necessary or in general may be too restrictive. * Fuzzy logic offers a practical way for designing nonlinear control systems. It achieves nonlinearly through piece-wise linear approximation. The basic building blocks of a fuzzy logical control system are set of fuzzy if-then (ie, fuzzy rule based models) that approximate a functional mapping. © Fuzzy logic provides an inference morphology that enables approximate human reasoning capabilities to be applied to knowledge-based systems. The theory of fuzzy logic provides a mathematical strength to capture the uncertainties associated with human cognitive processes, such as thinking and reasoning. * Fuzzy systems are suitable for uncertain or approximate reasoning, especially for the system with a mathematical mode that is difficult to derive. a @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms 1-6 Introduction . Fuzzy logic allows decision making with estimated values under incomplete or uncertain information. * Fuzzy logic is viewed as a formal mathe representation of uncertainty. Unce: management of real systems: matical theory for the rtainty is crucial for the if you had to park your car Precisely in one place, it would not be possible. Instead, you work within, say, 10 cm tolerances. * The presence of uncertainty is the price you pay for handling a complex system. Nevertheless, fuzzy logic is a mathematical formalism, and a membership grade is a precise number. What's crucial to realize is that fuzzy logic is a logic of fuzziness, not a logic which is itself fuzzy. Q.6 List and explain in brief constituents of soft computing. Ans.: © The principal constituents, ie,, tools, techniques, of Soft Computing (SC) are Fuzzy Logic (FL), Neural Networks (NN), Support Vector Machines (SVM), Evolutionary Computation (EC), and Machine Learning (ML) and Probabilistic Reasoning (PR). Fuzzy theory plays a leading role in soft computing and this stems from the fact that human reasoning is not crisp and admits degrees. Genetic algorithms © Genetic algorithms are inspired by Darwin's theory of natural evolution. In the natural world, organisms that are poorly suited for an environment die off, while those well-suited, prosper. Genetic algorithms search the space of individuals for good candidates. The chance of an individual's being selected is proportional to the amount by which its fitness is greater or less than its competitors’ fitness. Neural Networks © Neural networks consists of many number of simple elements *(meurons) connected between them in system. Whole system is Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithma 1+ _ Introducy,,, able to solve of complex tasks and to Jearn for it like 4 Natur, brain. e For user NN is black box with Input vector (source data, arg Output vector (result) Probabilistic reasoning © Uncertainty is described by probabilities. Probability may be for simulation of fuzziness. Us ¢ Relations between events are described a3 condition, or probabilities of transitic; probabilities (Markovian process). probabilities (Bayesian ne Q.7 Difference between digital computer and neural network. Ans. : Digital computer Neural network Deductive reasoning : We Inductive reasoning : Given input apply known rules to input | and output data (training data to produce output. examples), we construct the rules. Fault tolerant, redundancy and transistor goes and it no sharing of responsibilities. longer works. 2. | Computation is centralized, | Computation is collective, synchronous and serial. asynchronous and parallel. 3. | Memory is packetted, Memory is distributed, literally stored and location | internalized and content addressable. addressable. 4. | Not fault tolerant. One | Fast. Measured in millionths Slow. Measured in thousandths of| of a second. a second. Exact. Inexact. @ Less than PHOTOCOPY PriceIntroduction | Dynamic co _ nnectivity Applicable if well defined | Applicable MF rules ee ” rules with precise input data OF complicated or if data is none oF partial tlons of Soft Computing Q.8 What type of problem Ie solved by vy using soft computing ? Ans. : * Soft computing te: chniques have become one of tools that can Provide Promising Practice and reasonable solution 1. Feature Selection : selection, also known variable subset selecti relevant features for u: ¢ In machine learning and statistics, feature as variable selection, attribute selection or on, is the process of selecting a subset of ‘se in model construction. * Feature selection techniques are a subset of the more general field of feature extraction. Feature extraction creates new features from functions of the original features, returns a subset of the features, 2. Image processing: e In imaging science, image processing is any form of signal processing for which the input is an image, such as a photograph or video frame; the output of image processing may be either an image or a set of characteristics or parameters related to the image. whereas feature selection * Most image-processing techniques involve treating the image as a two-dimensional signal and applying standard signal Processing techniques to it. 3. Medical Diagnosis : ¢ Medical diagnosis refers both to the Process of attempting to determine or identify a possible disease and to the opinion reached by this process. From the Point of view of statistics the diagnostic procedure involves classification tests. 4, Pattern Recognition : « Pattern recognition generally aim to Provide a reasonable answer for all possible inputs and to perform @ Less than PHOTOCOPY Price4 Webrda sy Saft Computing tz Optimization Algor items 1-9 - = nyutn, taking WG OOITTE Hey inant likely" matebing, of the bnyauty, Ord L statintieal yariahon / jn studied in many bel and ethology, Cuyense e Pattern recog nihen hiatry, puter we lence psychology, P traffie flow and com Ihyent devices oe As inte connect wenn and 5, Smart Instrumentation ¢ allenge is to “ms. A major issue is the growing F with no single standard « The challenge is to intelligently connect smart inotramentahon At the that devices can communicate across multiple prots)s time, increases in the volume and importance A data robustness Of systers is ubiquitous the ¢ through smart sy" ureter +4 communication protocols, same means that privacy, security and paramount. Q.9 Explain shortest path algorithm. Ans.:¢ In many applications such as transportation, rou communications, economical, and s0 on, graphs emerge naturally as a mathematical model of the observed real world system. Indeed, many problems can be reformulated as a quest for a path between two nodes in a graph which is optimal in the sense of a number of preset criteria. © The object is to find a path of minimum cost or length fom a specified source node s to another specified sink node t, assuming that each arc (i, j) € A has an associated cost (or length) ij. ¢ Directed graph G = (V, A) where V is a set of nodes, A is a set of links. cj; is a cost associated with each arc(i, j) and Source node : node 1 Destination node : node n ¢ Indicator variable : xe fh if link (i, j) is included in the path uy 0, otherwise z DECODE) @ Less than PHOTOCOPY Price‘oft Computing & Introduction ° Fuzzy logic and fuzzy graph theor The basic idea bel arting at one point there until the destin, Y provides proper tools to use hind path finding is searching a , and exploring adjace ation node is reached, graph, s t nodes from * A path finding algorithm for transit handle the special characteristics of transit networks such as city emergency handling and drive guiding system, in where the optimal paths have to be found, network is proposed to In Dijkstra’s algorithm the weighted directe denote the se the graph is input of the algorithm consists of a ‘d_graph G and a source vertexs in Graph. Let's of all vertices in the graph G as V. Each edge of an ordered pair of vertices (u, v) representing a connection from vertex u to vertex v. © The set of all edges is denoted E. Weights of edges are given by a weight function w : E —~ [0, ©]; therefore w (u, v) is the non-negative cost of moving from vertex u to vertex v. The cost of an edge can be thought of as the distance between those two vertices. * The cost of a path between two vertices is the sum of costs of the edges in that path. For a given pair of vertices s and t in V, the algorithm finds the path from s to t with lowest cost (ie. the shortest path). It can also be used for finding costs of shortest paths from a single vertex s to all other vertices in the graph. * In a fuzzy graph G, let Abe the set of all paths from vertex a to vertex b and the fuzzy length of a path be Iq. ‘ 1, = Length (Q) = wiytwot.twi= DY wy, Qer eked * In the above equation, e, are edges of graph G. A fuzzy set S on 4 with memberships o, is the fuzzy set of shortest paths : 95 (Q) = min Wig 1) and how many solutions are just wasteful. 2. New offspring's are created by mutation. Each offspring solution is assessed by computing its fitness. Q.11 What is artificial neural network ? Explain task performed by ANN. Explain characteristics of ANN. Ans.: ¢ Artificial Neural Network (ANN) is a computational system inspired by the structure, processing method, learning ability of a biological brain. An artificial neural network is composed of many artificial neurons that are linked together according to specific network architecture. The objective of the neural network is to transform the inputs into meaningful outputs. a . (BECODEY @ Less than PHOTOCOPY PriceHida Algurithns 1-13 Introduction e ANN Pace Ga the pattern of inputs presented es for storing d ‘obi Wot execute programmed instructions; they Tespond jy, { to it. There are also No ata. Instead, information, —— ‘state’ the ne (s sautained in the overall activation ‘state ae whee etwor! a cl Kuwivledge! is thus represented by the a connor is r its individu ; quite literally more than the sum of its indivi Sey ale teiory adadre: © Fig Q1L1 shows artificial neural network. Input layer Input Input. Hidden layer Fig. Q.11.1 Artificial neural network «Elements of ANN are processing units, topology and learning algorithm. « Tasks to be solved by artificial neural networks : 1. Controlling the movements of a robot based on self-perception and other ‘information;. 2. Deciding the category of potential food items in an artificial world; ~ 3. Recognizing a visual object; 4. Predicting where a moving object goes, when a robot wants to catch it. © Characteristics of Artificial Neural Networks 1, Large number of very simple Processing neuron-like processing elements. 2. Large number of weighted connections between the elements. 3. Distributed representation of knowledge over the connections. 4. Knowledge is acquired by network through a learning process. > @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms 1-14 Introduct: IR - Introduction Q.12 List out the strength network. and weakness of artificial neural Ans. : Strength : 1. The greatest power of Neural Networks with a finite number of hidde: continuous function to is that it is endowed n units, can yet approximate any ; any desired degree of accuracy. This has been commonly referred to as the property of universal approximate. : 2. No prior knowledge of the d for implementing NN. 3. Problem of mode jata generating process is needed 1 misspecification does not occur. In case of NN since no Specifications are used as the network merely learns the hidden relationship in the data. 5. Adaptive learning : An ability to learn how to do tasks based on the data given for training or initial experience. 6. Self-Organisation : An ANN can create its own organisation or representation of the information it receives during learning time. Weakness : 1. The addition of too many hidden units incites the problem of over fitting the data. 2. The construction of the NN model can be a time consuming process. Q.13 Draw and explain architecture of neural network. Ans.: ¢ The architecture of the neural network refers to the arrangement of the connection between neurons, processing element, number of layers, and the flow of signal in the neural network. ¢ There are mainly two category of neural network architecture : a. Feed-forward : b. Feedback (recurrent) neural networks. '@ Less than PHOTOCOPY PriceSoft Computing 1. . Inputs f+ Optimization Algoritiona 1-15 Introducti, Architecture and Learning Rule In late 1950s, Frank Rosenblatt introduced a network COMPoseq of the units that) were enhanced version of McCulloch-Pjt,, Threshold Logic Unit (TLU) model. Rosenblatt’s model of neuron, a perceptron, was the result oy merger between two concepts from the 1940s, McCulloch-Pity, model of an artificial neuron and Hebbian learning rule o adjusting weights. In addition to the variable weight values, the perceptron mode added an extra input that represents bias. Thus, the modifies equation is now as follows : N sum = SI, Wit b ist where b represents the bias value. Fig. Q.13.1 shows a typical perception setup for pattem recognition applications, in which visual patterns are represented as matrices of elements between 0 and 1. Activation function (more on this later) Output “Fig. Q.13.1 . First layer act as a set of feature detectors that are hardwired to the input signals to detect specific features. . Second layer ie. output layer takes the outputs of the feature detectors in the first layer and classifies the given input pattern SS = @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms 1-16 Introduction eLearning is initiated by making adjustments to the re! connection strengths and a threshold value 6 evant e Here we consider only two class problem, Here output layer usually has only a single node. For an n-class problem (n> 3), the output layer usually has n-nodes, © ach corresponding to a clas and the output node with the largest value indicates which class the input vector belongs to. ¢ In the first stage, the linear combination of inputs is calculated Each value of input array is associated with its weight value, which is normally between 0 and 1. Also, the summation function often takes an extra input value Theta with weight value of 1 to represent threshold or bias of a neuron. a. The term x; is referred to as active or excitatory if its value is 1. b. If the value is 0 then it is inactive, c. If the value is -1 then it is inhibitory. ¢ The output unit is a linear threshold element with a threshold value 0: 0 = £(S7. wi % -8) = £(Dh, xt wo)Wo SG -*(hiw xi), Xp #1 where wj; is a modifiable weight associated with an incoming signal x;. f) ' oO Fig. Q.13.2 shows the bias term wg. Xo xX; Ww, x 2 Wr .o= 3 Xp ° Xg W3 8 Fig. Q.13.2 Bias term wo @ Less than PHOTOCOPY Price7 Computing & Optimisation Alyoritune fet Mo odue,, # The tunetton y= IQ) describes relationship, an Input outpy mapping front to y @ The equation (Q 111), the (0) in the activation fanetion Of the Pereeptron and it is typically either a alga N function npn(y 4, step function step() : ja ix 0, san) = {1 otherwise ; JL ityso, ww (Q3 sey ) lo othery 4 ¢ The sunvot-product value is then passed into the second stage to Perform the activation function which generates the output from the neuron. The activation function “squashes” the amplitude of the output in the range of (0, 1] or [-1, 1] alternately, The behavior of the activation function will describe the characteristics of an artificial neuron model. * The basic learning algorithm for a single layer perceptron repeats the following steps until the weights converge : 1. Select an input vector x from the training d 2. If the perceptron gives an incorre connection weights w; according to lata set. ct response, modify all Aw; = nti X; Where t; is a target output and nis a learning state. Q.14 Explain application of neural network. Ans. : Neural network applications can be Brouped in following categories : 1. Clustering : A clustering algorithm explores the similarity ms and places similar patterns in a cluster. Best known applications include data compression and data mining. 2. Classification/Pattern recognition : The task of pattern Tecognition is to assign an input pattern (like handwritten symbol) to one of many classes. This category includes algorithmic implementations such as associative memory. — @ Less than PHOTOCOPY PriceSoft Computing & Optimization Aly: 3. Function approximation ; is to find an estimate of noise. Various engineering function approximation. “ 4. Prediction/Dynamical systems : future values of a time-sequenced 4 significant impact on decision support differs from function approximation by con Q.15 List the advantages of neural network. Ans.: The advantages of neural networks are due to its ad and generalization ability. a) Neural networks are adaptive methods that can learn without any prior assumption of the under data. b) Neural network, namely the feed forward multilayer perception and radial basis function network have been proven to be universal functional approximations. c) Neural networks are non-linear model with good generalization ability. Q.16 What is hybrid system ? List and explain classification of hybrid system. Ans.: ¢ In hybrid system, more than one technology is used. Integrated architectures for machine learning have been shown to provide performance improvements over single representation architectures. e The combination of knowledge based systems, neural networks and evolutionary computation forms the core of an emerging approach to building hybrid intelligent systems capable of reasoning and learning in an uncertain and imprecise environment. eIn recent years multiple module integrated machine learning systems have been developed to overcome the limitations inherent in single component systems. Integrations of neural DECODE) @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms 1-19 Introduction networks, fuzzy logic and global optimization algorithms have received considerable attention but increasing attention is bein, Paid to integrations with case based reasoning and ry), induction. * Classification of hybrid systems : 1. Sequential hybrid systems 2. Auxiliary hybrid systems 3. Embedded hybrid systems Sequential Hybrid Systems © Sequential hybrid system uses Pipeline concept. The output of one technology is used as input of other technology. * Genetic algorithm Preprocessor is an example of sequential hybrid system. It gives optimal parameters for different instances of a problem and give it to preprocessed data set to an neural network. ° Fig. Q.16.1 shows sequential hybrid system. Fig. Q.16.1 * It creates problem while combinin; ig two technologies, Auxillary Hybrid System . @ Less than PHOTOCOPY PriceIntroduction Second technology (Subroutine) * Neuro-genetic system is an example of auxiliary hybrid system. Embedded Hybrid System ¢ Two technologies are integrated such a way that they appear intertwined. Embedded system is hybrid of hardware and software. Neuro-Fuzzy system is an example of embedded hybrid system. « Embedded systems specifications usually impose very tight constraints with respect to cost, response time and power consumption, among other characteristics. They often include software, hardware and interfacing subsystems. Furthermore, the design of these kinds of systems often requires concurrent optimization of several design objectives, which are conflicting in most of the cases. Q.17 Compare neural processing and fuzzy processing. Ans. : Neural processing Fuzzy processing It tries to incorporate human _| Fuzzy logic allows making definite thinking process to solve decisions based on imprecise or problems without ambiguous data. mathematically modeling. @ Less than PHOTOCOPY PriceSoft Computing ning is start from scratch. ed. Learning algorithm is _ us Behavior is like black box. Neural networks are good at recognizing patterns, they are not good at explaining how they reach their decisions. & Optimization Algorithms 1-21 Introd, Soy Priori knowledge is required, Learning algorithm is not used, Simple to implement . Fuzzy logic can reason with im: information, are good at explain; their decisions but they cannot automatically acquire the rules use to make those decisions, & they END. e @ Less than PHOTOCOPY PriceUNID wd | 2 | we s * ‘~! Lonte | 2.0 Baste op Q41 What te examples, want by fussy loule « en? Musteate tt with Proper Ante Bury net theory wan developed by Lotti A Zadeh, Professor for computer selence at the Univernity of California in Berkeley, to provide a mathematical tool for dealing with the concepts used in natural lang * Fuzzy logic is not logic that in fuzzy, describe fuzziness, Fuzzy logic is the that calibrate vagueness, but logic that is used to theory of fuzzy sets, sets Fuzzy Logie is basically a multi-valued Mte values to be defined evaluations. logic that allows between conventional Fuzzy logic is a form of many-valued logic in which the truth values of variables may be any real number between 0 and 1, intermed Fuzzy logic is an extension of Boole environments where there is no ab: uncertainty, an logic used to describe solute truth and there is Fuzzy logic was specifically designed to mathemati ally represent uncertainty and vagueness and to provide formalized tools for dealing with the imprecision intrinsic to many problems. Fuzzy logic provides an inference morphology that enables approximate human reasoning capabilities to be applied to knowledge-based systems. The theory of fuzzy logic provides a ceeeeeeemmmnmeeeeeey @-1)2-2 Fuzzy Sets andy 4) thematical strength to capture the uncertainties associated Wit, mathematical streny, H I b cognitive processes, such as thinking and reasoning human cog Fuzzy systems ate knowledge-based or rule-baned systems, 1, @ Furey systems are ‘ a heart of a fuzzy system is a knowledge base consinting, of so-called fuzzy IF-THEN rules, FL is a problem-solving control system methodology that lena, itself to implementation in s stems ranging from simple, gm, embedded micro-controllers to large, networked, multi-channg) PC or workstation-based data acquisi ion and control systems, The conventional approaches to knowledge representation tac; the means for representing the meaning of fuzzy concepts. Ag a consequence, the approaches based on first. order classical probability theory do not conceptual framework for dealing wi common sense knowledge, both lexically imprecise and What is Fuzziness ? logic ang provide an appropriate th the representation of since such knowledge is by its nature non-categorical. * For example, if pressure takes values between 0 and 50, one might label th e range 20 to 30 as medium Pressure. becomes a member. * Fuzziness is not vague and multi-valued logic. * Fuzzy logic may be considered as an extension of multi-valued logic but they are somewhat different, Multi-valued logic is still based on exact reasoning whereas fuzzy logic is approximate reasoning. Less than PHOTOCOPY PriceSoft Computing & Opti Suton Algorithms 2. 3 Fuzzy Sets and Logic $a ee ae bogie mathematical strep th to ¢ roe apture the human cognitive ; Stel ag Uncertainties associated with * thinking and re. asoning @ The conventional 'PPreaches 10. knowledje " representation lack Hlation the meaning of furzy concepts the means for repr o The development of fuzzy logic was Motivated in J. d for al frame work which c and lexi by the ne large measure 4 conceptu, an address the issue of une ertainty al imprecision : Q2 List out the characteriatics feature Ans.: ¢ Essential characteristics of fuzzy logic : 1. In fuzzy logic, exact Teasoning is viewed as a limiting case of approximate reasoning. "of fuzzy logic, 2. Everything is a matter of de; 3. Knowledge — is equivalently, Bree in fuzzy logic, interpreted a collection of fuzzy constraint ona 4 Inference is viewed as a process of Propagation of elastic constraints, elastic or, collection of variables. 5. Any logical system can be fuzzitied, e There are two main characteristics of fuzzy systems that give them bette: * performance for specific applications. 1. Fuzzy systems are suitable for uncertain or approximate reasoning, especially for the system with a mathematical model that is difficult to derive. 2. Fuzzy logic allows decision making with estimated values under incomplete or uncertain information. * Fuzzy systems are suitable for uncertain or approximate reasoning, especially for the system with a mathematical model that is difficult to derive. Fuzzy logic allows decision making with estimated values under incomplete or uncertain information. * Fuzzy logic is based on the idea that all things admit of degrees. For example, temperature, height, speed, distance and beauty all come on a sliding scale. The motor is running really hot. Ram is a very tall guy. (ECODE) @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms 2-4 Fuzzy Sets ang y ° Bi: Q.3 List the merits and demerits of fuzzy logic. Ans.: 0 . Allows the use of vague linguistic terms in the rules, a erits of fuzzy logic - Modularity : Rules can be added and removed as needed, Parallel execution of rules : Output calculated once at end a cycle. - Fuzzy logic inherently accounts for noise in the dat, because it extracts trends, not precise values, . Algorithms in fuzzy logic are cast in the sam language used in day-to-day conversation which makes fuzzy logic predictions _ easily interpretable. . Fuzzy logic algorithms are computationally _ efficient and can be scaled to include an unlimited number oj components. Demerits of fuzzy logic a 2. 3. 4 Difficult to estimate membership function. There are many ways of interpreting fuzzy rules. Computational cost : more computations involved. Rules need to be expressive and needs to be accurate. 2. Fuzzy Sets and Crisp Sets Q.4 Explain following fuzzy set operations with example : a. Empty fuzzy set b. Product fuzzy set c. Normal fuzzy set Ans.: ¢ Let X be a nonempty set. A fuzzy set A in X i characterized by its membership function Ba : X10, 1] and pq (x) is interpreted as the degree of membership of element’ in fuzzy set A for each x € X. © It is clear that A is completely determined by the set of tuples A = {(aHa (u))lue X} ‘ @ Less than PHOTOCOPY PriceSoft Computing & Optimis Sof Computing & Optimization Algonthms 2-5 Fuzzy Sets and Logic Frequently we will write A(x) instead of js, (x) The family of all tuzzy sets in X is denoted by F (x). A amily H If N= (qs Xn] is a finite set and A is a fuzzy set in X then we often use the notation ° AT MX tt eg xp where the term it xy, i = L...n signifies that 1; is the grade of membership of x; in A and the plus sign represents the union. i) Normal fuzzy set : A fuzzy subset A of a classical set X is called normal if there exists an xX such that A(x) = 1. otherwise A is subnormal. ii) Product fuzzy Set : Let A and B be the fuzzy sets in X and Y respectively. The cartesion product of A and B denoted by A x B, is a fuzzy set in the product space X x Y with the membership function Maxp(%y) = mini, (x)Hp (x) Cartesian co-product A+B is a fuzzy set with the membership function Hass (%y) = max, (x)Hp (x) iii)Empty fuzzy set : The empty fuzzy subset of X is defined as the fuzzy subset 6 of X such that (x) = 0 for each x € X It is easy to see that 6c A holds for any fuzzy subset A of X. Q.5 What is the alpha-cut method for discrete fuzzy sets ? Ans. : An a cut of a fuzzy set A is a crisp set “ A that contains all the elements in X that have membership value in A strictly greater than or equal to . fA — {xl A(x) 20} © A strong o-cut of a fuzzy set A is crisp set @ A that contains all the elements in X that have membership value in A strictly greater than o. a+ A - {x A(x) > a than PHOTOCOPY Pricetion Algorithms 2-6 Fuzzy Sets and Ly, Soft Computing & Optimt n extracted crisp set The value ., ie. ve [0,1]. The o-cut of , the set of all element a fuzzy set into a © cuts reduce represents a membership degree, set Ais the crisp set (A-@) Le, 20. furzy whose membership degress in A are Q.6 “Behavior of fuzzy logic Ie deterministic’? Justify. viewed as a super It adds degrees betwe etwe set of Boolean Ans: ¢ Fuzzy logic can b n the absolute logic, as a mult-valued logic. truth and absolute false to cover partial truth in b ein simple terms, fuzzy logic involves classifying objects anc functions into fuzzy sets which could be given linguistic phrases neither exact, nor absolutel, et is a form of reasoning that is : phrases which do n> inexact. For example, too hot, little slow, give the idea of absolute, but a fuzzy estimate. ough for a person from tt ght find the heat unbearabl: « While 33 degrees might be warm ¢1 equator, someone from the arctic mi, or too hot. e It is not possible to classify them into strict sets with define boundaries which leads to the idea of fuzziness. «It has basically evolved from predicate logic, though many fo called t-norm fuzzy logics do exist in propositional logic too. generally has an object and a predicate. For example : in the sentence, “Plato is a man", Plato’ is t object, and "is a man" is the predicate. But an important po about fuzzy logic is that it is deterministic and time-variant. e A few salient points on fuzzy logic are : 1. Exact reasoning or precise values is the extreme or limitir. case of approximate reasoning. 2. Any system which works on logic can be fuzzified a everything would be a matter of degrees. 3. Knowledge is a collection of fuz: i ee Zy constraints on a group’ EF = @ECODE) @ Less than PHOTOCOPY PriceSoft Computing & Op; om Alcorithme yg embership functions ? _ Q.7 What ts m, Fuzzy Sets and Logic = ee function Is repr, How fussy membership Ans. : © Fuzzy sep is any set different degree Of me interval [0, 1] e Fuzzy sets are ch. hat allows ite mlserahip Members to have falled Membership function in the atacterized py Memberships Pton that » °8 place tion + The membership function seta number, A(X), in the A(x), represents the membership functions set has a graphic fre al desc, Function A furry “Presses how th THis graphical J Yoone to another tak 1@ transition mda membership funy mai eaeription iq PSSIRNS to each element x in a furry Closed unit interval [9,1] degree of membe are functions of th The number, tship of x in A. Fence, Ne form: A: X -+ [0, 1] e In the case of Crisp sets, the Members of a set are either the set, with membership degre value one being the degree of m © Therefore, Crisp Sets ¢ Fuzzy are special cases of fuzzy sets, out of € Of zero, or in the set, with the embership, Sets or in other words, crisp sets ¢ An element can be in the set with a degree of m, out of the set with a degree of membership, © Hence, crisp sets are @ special case, or a subset, of fuzzy sets, where elements are allowed a membership degree of 100 % or 9 % but nothing in between. e There are four ways of representing fuzzy membership functions, namely, a. Graphical representation, b. Tabular and list Tepresentation, c. Geometric representation, d. Analytic representation ° Tabular and list representations are used for finite sets. ¢The third method of representation is the geometric representation and is also used for representing finite sets. @ Less than PHOTOCOPY Priceapaitinng te Optimisation Algorithina 2-8 Fuzzy Sets and Lo, — Analytical representation is another alternative to graphic, Ming, infinite sets, @f. a set of rey ne vey station iy repre rumibera 2 The xpectiication of membership: functions is subjective, whict means that the membership funetions specified for the same concept by different persona may vary conniderably. Inn methoda employed for membership value Ana. t Memberahip value assignments methods : 1. Intuition 2. Ife A. Rank ordering 4. Neural networks e Intuition ; Intuition involves contextual and semantic knowleds: «; it can also involve linguistic truth values abou. Age. « Inference ; In the inference method we use knowledge to perfor: deductive reasoning. Let A, B, and C be the inner angles of ; triangle, in the order A 2B 2 C 20, and let U be the univer, of triangles, Le about an i this know U = (A,B, QIAZB2C20;A+B+C} = 180° Rank ordering : Assessing preferences by a single individual, a committee, a poll, and other opinion methods can be used to assign membership values to a fuzzy variable. Preference is determined by pair-wise comparisons, and these determine the ordering of the membership. ¢ Neural Network : It is a.technique that seeks to build a intelligent program using models that simulate the working network of the neurons in the human brain. In a global sense a neuron receives a set of input pulses and sends out anothe pulse that is a function of the input pulses. Neural system solve problems by adapting to the nature of the data(signa! they receive. @ Less than PHOTOCOPY PriceSoft Computing & Optimization Alzorithms 2 Fuzzy Sets and Logic Once the Neural performing Satisfact, Of any other data p network is t ‘orily, it can be OINts in the * A complete mapping of the me in the different fuzzy classes overlap of the different classes. Tained and verified to be used to find the membership two fuzzy classes mbership of different data Points can be derived to determine the 2.3: Fuzzy Set Theory and Operations Q.9 Explain fuzzy logic Operations, Ans.: ¢ Union, intersection, and complement are the most basic operations on classical sets. On the basis of these three operations, a number of identities can be established. Corresponding to the ordinary set Operations of ‘union, intersection, and complement, fuzzy sets have similar operations, 1. Containment or subset : Fuzzy set A is contained in fuzzy set B (or, equivalently, A is a subset of B, or A is smaller than or equal to B, A c B) if and only if 1.4 (x) Sug(x) for all x. AcBeoyp A Sg A is contained in B Membership grades Fig. Q.9.1 Concept of Ac B 2. Union (disjunction) : The union of two fuzzy sets A and nee fuzzy set C, written as C= AUBorC=AOR B, whose related to those of A and B by w(x) = max(ug (x), BBX). Cm AU B & u(x) = max (Hg (x) B(X) =a OV BBD ——————— (DECODE) @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms —2- 10 Fuzzy Seta and Loyi, 3. Intersection (conjunction) : The Intersection of two fuzzy mets A and B is a fuzzy set C, written as C = A) Bor C= A AND &, whose MF is related to those of A and B by #0) = mink, (), HRD) CHA OB © te CX) = min, OX), 4. Complement (negation) + The denoted by A or sO) HACIA Bply) complement of fuzzy set A, NOT A, is defined as pa (x) = 1 - pA) 5. Cartesian product and co-product : Let A and B be the furzy sets in X and Y respectively. The cartenion product of A and & denoted by A x B, is a fuzzy set in the product space X + with the membership function Baxs & y) = mint, C0, b pCO). Cartesian co-product A + B is a fuzzy set with the members}, function Bas By) * max (x), BB OO): Fig. Q.9.2 shows the basic three operation on two fuzzy set A 2:, B. a) Fuzzy sets A and B b) Fuzzy set "not A” 0s ° c) Fuzzy set"A OR B” d) Fuzzy sets “A and B” —___—_— 4 1 ; 05 0s ° ° Fig. Q.9.2 O; ation on fuzzy sets =soft Computing & Optimization Algorithms 2-11 Fuzzy Sets and Logic 0.10 Consider two fuzzy sete A and B (0.5, 0.7, 0.2/0.4 Bn [err ge Gee a perform the following operating on fuzzy sets ) AUB 1) ACB tt) Compement of fuzzy set A fv) Difference (3) v) Algebraic sum of given fuzzy sets. vi) Bounded eum of the given fuzzy set. vil) Algebraic product of the given fuzzy sets. vill) ACB i) AUB Ans: i) AUB ii) iii) Complement of fuzzy set A: 1-p, (x) - 0 0.5 0.7 0.8 2s ES a ‘ A . iv) Difference 3/7 ANB = A-(AMB) v) Algebraic sum = @ECODES @ Loss than PHOTOCOPY Price2 Veiga deg & Qevevsatn Algorithms 2-12 Ni SQhatad sua = min ] x = (10.7 0.3 *2| n) AUB= Calculate, B = 0 . . 6 AUB = {3+9+% } Q.11 Consider two fuzzy sets X and Y, find Complement, Union, Intersection, Difference : x=? @ecopey @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algori 2-13 Fuzzy Seta and Logic set y = {9:7 0:2 0.4 0.2 0.8 0.8 0.7| [2°37 45 6°77 Bf 2) Untion = . {3.28 0.6 0.8 0.3 0.7 0.8} 3) Intersection : XaY 4) Difference (XY) = XY = X-(XnyY) {F384 oo FI Fuzzy and Crisp Relations, Fuzzy to Crisp Conversion Q.412 Define and explain classical relations and. fuzzy relations. ‘Ans ¢ Relations are the way in which two or more things are connected. In natural language, relations are kinds of links existing between objects such as "mother of", “neighbor of", "part of", etc. * Mathematically, a relation implies the presence of an association between elements of different sets (at least two fuzzy sets). Binary relation is a relation between two sets. Tertiary relation is a relation between three sets and n-ary relation is a relation between n elements. © Classical relations are structure that represents the presence or absence of correlation or interaction among various sets. Two degrees of relationship in a crisp relations : Completely related and Not related. @ Less than PHOTOCOPY Price eyesSoft Computing & Optimization Algorithms 2 - 14 Fuzzy Sete and ton, n between two sets X and Y denoteg © A binary relation is a relatio lation can be expressed as by R(XY). A binary fuzzy rel ie. through statement 1. Linguistically, red pairs By listing the set of all orde Directed graph Matrix relation R Table anep x ty Fig. Q.12.1 ° A fuzzy relation is based on the philosophy that everything is related to some extend or unrelated. Unlike crisp relations, the strength of the relation between ordered pair of two universes is not measured with the characteristic function, but rather witha membership function expression expressing various degree of strength of the relation on the unit interval [0,1]. © Crisp relations and fuzzy relations can be defined in terms of subsets. «A binary fuzzy relation is a fuzzy relation between two sets * and Y denoted by R(X,Y). A binary fuzzy relation ca" be expressed as a. By listing the set of all ordered pairs b. Directed graph c. Matrix relation R d. Table (Tabular form) @ Less than PHOTOCOPY PriceSoft Computing & Optimization Aluorttome 7. 45 4 id Nags a ti g Nuriy Sete und Vigie # Let X and Y be two univernen Of dina ourne, 1 ure, than R= & WM AYN yese 4 is a binary fuzzy relation in x» ” yn) y i y, e Puzzy relations represents the ntrenyth of ansiation betieen elements of the two nots, : Example : R= "x iy considerably larger than ROG Y) = Relation between nets X and ¥ Ri y) = Membership function for the relation R (X,Y) R(% Y) = Ri NIGYN KH yEeKe Hy} 0 furxcy Roy y) = (x~y)/(10- yy for y11y o Let X = Y= Re and R = "y is much greater than x" . The member function of the fuzzy relat ition R can be subjectively defined as yrx, URG&Y) = {x+ye2 yes 0 ifysx Q.13 What are fuzzy relations? Explain following operations on fuzzy relations. . 4) Intersection i) Containment Ans.: © Union, intersection, and complement are the most basic operations on classical sets. On the basis of these three operations, a number of identities can be established. * Corresponding to the ordinary set operations of union, intersection, and complement, fuzzy sets have similar operations. Intersection (conjunction) : The intersection of two fuzzy sets A and B is a fuzzy set C, written as C = ANB or C= A And B, whose MF is related to those of A and B by Bo) = min 4 (0,1 ()) Cm AOB © Hc(x) = min 4 (X)4g ()) =BaACAKB(X) @ Less than PHOTOCOPY Priceei ization Algorithms" ‘uzzy Sets and}... sft Computing & Optinia! tion Algorith 2-16 ft e i i in fuzzy set B i . Fuzzy set A is contained in re en ina onbeet of B, or A is smaller than or equal to (equivalently, (&) Shy (%) for all x. B, Ac B) if and only if ta AGB? Ita Sip A is contained in B Membership grades Fig. Q.13.1 Concept of AC B Q.14 What is T-norm, T-conorm? Explain order of T-norms. Ans. : 1, T-norm « A tnorm (triangular norm) is a kind of binary operation used in the framework of probabilistic metric spaces and in multi-valued logic, specifically in fuzzy logic. A t-norm generalizes intersection in a lattice and conjunction in logic. * Fuzzy intersection of two fuzzy sets can be specified in a more general way by a binary operation on the unit interval, ie, a function of the form : . T : [0, 1] x [0, 1] — [0, 1] * In order for a functionTto qualify as a fuzzy intersection, it must have appropriate properties. Functions known - ast-norms (triangular norms) posses the Properties re-quired for the intersection. Similarly, functions calledt-conorms can be used fot the fuzzy union. ° Atnorm T is a binary operation on. the unit interval that satisfies . at least the following axioms for all | a,b; ce[ 0, 1}. t et @ Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms 2-17 orms have follo, Ng order: tmin (@/b) S max O,a+b~1) < ay S min (a,b) M1) T&y)=Ty, x), ie, the t-norm is commutative, @ TTY 2-7 7 (y, 2)), ie, the t-norm is Associative (73) xSy=> T(x, zs Ty, 2), ie, the t-norm is Monotone, (4) T(x, 1) =x, ie, a neutral element exists, which is 1. e Some frequently used t-norms are 5 1. standard (Zadeh) intersection: T(a; b) = min (a; b) 2. algebraic product (probabilistic intersection) : T(a; b) = ab . 3. Likewise (bold) intersection : T(a; b) = max (0; a + b — 1) ¢ Although continuous t-norms are most commonly used in fuzzy logics and have an elegant representation 2. T-conorm °A tconorm S is a binary operation on the unit interval that satisfies at least the following axioms for all a; b; c € [0; 1] ¢ Some frequently used t-conorms are: 1. standard (Zadeh) union: S(a; b) = max(a; b); 2. algebraic sum (probabilistic union) : S(a; b) = a + b - ab; 3. Like wise (bold) union: S(a; b) = min(1; a + b) The minimum is the biggest t-norm and maximum is the smallest t-conorm. @ Less than PHOTOCOPY Pricethine 28 Purr Sets ang) Opetnaa toon Aly Sot Come ting © ————— tere : [ e in Furry Logie nterferenc ble diagram, explain construction B nee eyete Qs Wiehe working of fu 7 . ALS syste ay Interenes Syatent (PIS) da a syater that yy puts to outputs, FIS uses 4 collection instead of Boolean logic Ans: ¢ A Pur fussy set theory to map it] yo membership finetions and rules, t reason about data “1p the service is good, fe For instance 5 nen if the food is not excelte, the tye will be genes”. « FIS are used to solve decision problems, ie. to make a decisi,, and act acconlingly. Structure of a fuzzy Inference system In general a fuzzy inference system consists of four modules : Fuzzification module 2. Knowledge base 3. Inference engine 4. . . Defuzzification module Fig. Q.15.1 shows block diagram of FIS. Crisp value Crisp value Fig. Q.15.1 Block diagram of FIS 1. Fuzzification module : Transforms the system inputs, which ¢ crisp numbers, into fuzzy sets. This is di ‘applyi fuzzification function. ens ti eres 2. Knowledge base : Stores IF-THEN rules provided by experts. @ Less than PHOTOCOPY PriceComputing & Optimizany, Soft 1g & Op: mization Algorithms 2-19 Fuzzy Sets and Logic . Inference engine : s; 3 gine : Simulates the human reasoning process by nce on the inputs and s IF-THEN rules. efuzzificati 4D on module : Transforms the fuzzy set obtained by the inference engine j ; fuzzy output cet ig 2 8 SSP Value. In defuzzification, the commonly used. te is (Converted to a crisp number. Some methods. chniques are the centroid and maximum Why should we use fuzzy Inference systems? 1. Fuzzy logic does not solve new Problems. It uses new methods to solve everyday problems, 2. Mathematical concepts within fuzzy Teasoning are very simple. 3. Fuzzy logic is flexible : It is easy to modify a FIS just by adding or deleting rules. There is no need to create a new FIS from scratch. 4. Fuzzy logic allows imprecise data (it does NOT work with uncertainty) : It handles elements in a fuzzy set, ie. membership values. For instance, fuzzy logic works with ‘He is tall to the + degree 0.8' instead of "He is 180 cm tall’. 5. Fuzzy logic is built on top of the knowledge of experts : It relies on the know-how of the ones who understand the system. 6. Fuzzy logic can be blended with other classic control techniques. Q.16 Mention the application area of fuzzy logic. Ans.: Fuzzy logic has rapidly become one of the most successful of today's technologies for developing sophisticated control systems. The reason for which is very simple. Fuzzy logic addresses such applications perfectly as it resembles human decision making with an ability to generate precise solutions from certain or approximate information. It fills an important gap in engineering design methods left vacant by purely mathematical approaches (e.g. linear control '@ Less than PHOTOCOPY PriceSoft Computing & Optimization Algonthms 2-20 Fuzzy Sets ang lon, ly logiccbased approaches (@ 8: expert syston. design), and pu in system design. While other approaches require accurate equations to Mode real-world behaviors, fuzzy design can accommod ate 4. ambiguites of real-world human language and logic. «It provides both an intuitive method for describing systems in human terms and automates the conversion of those System specifications into effective models. * Fuzzy ystem applications can be found in many fields of human activity, especially in control tasks of nonlinear and complex systems, where expert knowledge in the form of fuzzy rules js known. * They can be classified according to the tasks to be performed and the domain of application. * Also known as TSK fuzzy model. Takagi, Sugeno and Kang was Proposed generation of fuzzy rules from a given input-output data set. ° A TSK fuzzy rule is of the form : "If x is A and y is B then Z= f(x, y)” where A and B are fuzzy sets in the anteced lent, while z = f(x, y) is a crisp function in the consequent. * f(,.) is very often a polynomial function w.r.t. x and y elf £.,.) is a first order Polynomial, then the resulting fuzzy inference is called a first 01 der Sugeno fuzzy model. ° If £(,.) is a constant then it isa zero-order Sugeno fuzzy model (special case of Mamdani model). * Case of two rules with a first-order Sugeno fuzzy model. 1. Each rule has a crisp output. 2. Overall output is obtained vi ia weighted average. 3. No defuzzyfication required. @ Less than PHOTOCOPY Price4 N0n Ayurithinng ¥ Sets and Logic paanple Ly Stugle a Ottputinpur Sug, WOO fussy Model with three EN is medians then y TEN is Large the, aM" and", put curve is At8e" are nontuzey © Piece Wise son between fuz sets then the overall linear, Following Fig. Q.161 Vand nontuzzy rules of eample 1. (©) Overait VO curve for eriep rules =s ° 8 10 Membership yrades Fig. Q.16.1 * However, if we have smooth membership functions (fuzzy rules) the overall input-output curve becomes a smoother one. Example 2 : Two-input single output fuzzy model with 4 rules R1: If X is small and Y is small then z =~ xtytl j / | € 2 NN { — +—+ ° 0 20 30 40 50 60 70 80 a X= Age Fig. Q.17.1 Membership function of the term set T(age) Constructing MF for composite linguistic terms * Meaning of the linguistic terms young and old be defined by the following membership functions : pyoung(x) = bell(x, 20, 2, 0) 1 1+ (x/20)4 nold(x) = bell(x, 30, 3, 100) 1 1+ ((x- 100) / 30)® ° Where x is the age of a given person, with the interval [0, 100) as the universe of discourse. Now we construct for the following composite linguistic terms : 1. More or less old = DIL(old) = old®5 2. Not young and not old Linguistic variables and linguistic values : * By a linguistic variable we mean words or sentences in a_natur. example, Age is a linguistic vari rather than numerical, .e., youn; young, old, not very old and no! a variable whose values are ‘al or artificial language. For able if its values are linguistic '& Not young, very young, quite t very young, etc. = G@ECODEY @ Less than PHOTOCOPY Price& Optimi: soft COMPUTES piimization Algorithms 2-25 Fuzzy Sets and Logic suppose that X = “age.” Then we oS Sethe can define fuzzy sets "young," middle aged,” and “old” that are a y yous haracterized by . just as a variable can assume various can assume differe ‘middle aged” and "old" of "young," then we have forth for the other values. values, a linguistic variable nt linguistic values, such as "young," in this case. If "age" assumes the value the expression "age is young,” and so «The name of a linguistic variable is its label. The set of values that it can take is called its term set. Each value in the term set is a linguistic value or term defined over a universe. In summary : A linguistic variable takes a linguistic value, which is a fuzzy set defined on the universe. « Example : Let x be a linguistic variable labeled ' ‘Age’. Its term set T could be defined as T (age) = {very young, young, not very young, more or less old, old} «Each term is defined on the universe, for example the integers from 0 to 100 years. ¢ The support of a fuzzy set A is the set of all points x in X such that [, (x) > 0. * The core of a fuzzy set A is the set of all points x in X such that Ha (x) = 1. * A fuzzy set A is normal if its core is nonempty. In other words, we can always find at least a point x € X such that va) =1. ¢ A crossover point of a fuzzy set A is a point x € X at which Ha) = 0.5. * A fuzzy set whose support is a single point in X with p a@®=1 is called a fuzzy singleton. @ Less than PHOTOCOPY PriceSo? Come tay S Ortenzanon Alguntims 2-26 FUE Sety ang eS OF ; b Qt$ Explain furry Mt Then mul fry sets and furry sets operations are the Subjects logic. It Then nile statements are used to fory nents that compnse fuzzy logic. Mula, s also called fuzzy rule, fuzzy implication honal statement. It assume the form % If x is A then y is B B are hnguistic values defined by fuzzy sets , sscourse \ and Y respectively. * + Onen \ is Ais called the antecedent or premise while "y ig, Sed the consequence or conclusion. codent describes to what degree the rule applies, whj, sion assigns a fuzzy function to each of one or mo, variables. ; « If pressure is high, then volume is small. © If the road is slippery, then driving is dangerous. * If a tomato is red, then it is ripe. * If the speed is high, then apply the brake a little. Example : Speed and pressure of a steam engine can be expressed with the following linguistic conditional statement If Speed is SlowThen Pressure should be High. Graphically, this statement looks like as shown in Fig. Q.18.1. ° The fuzzy rule "If x is A then y is B" may be abbreviated as A — B and is interpreted as A x B. A fuzzy if then rule may k defined (Mamdani) as a binary fuzzy relation R on the produc space Xx Y. R= A—>B=AxB = fu, COHAMI&y) DECODE Se Less than PHOTOCOPY Price \RECODE)uting & Optimi: oft COMP ‘S ptimization Algorithms a» Tray Seta and Logie yooh Ayz Slow Ans yy Bye low Oy aban xa Speed Pras Fig. Q.18.1 If-then rule ‘ : A — B is interpreted as A ent; I A different formulas : fails B, then it can be written as . Material implication R=A >B ; =aA . propositional calculus vB R=A>B*7Au (A « Extended propositional calculus de R=A>B=(AAN-B B « Generalization of modus ponens Raaeta pry) = sup fe [Ha &) X csp Bly) and O B. Two ways to interpret "If x is A then y is BY: y (b) A entails B tation of a fuzzy rule A> B ations and various T-norm and : of qualified methods can be B. R can be viewed as (a) A coupled with B Fig. Q.18.2 Two Interpret Based on these two interpret T-conorm operators, a numbe! formulated to calculate the fuzzy rule A > ensional MF : a fuzzy set with a two dim © Less than PHOTOCOPY Price' } 2-28 Fuzzy Set, Soft Computing & Optimizati 'Y Sets ang Logie Bey) = FLA DM Bly) = fa, b) t Here f is called fuzzy implication function performs the ere f is ca task ip grades of x in A and y in B ing Of | transferring the members © those of (x y) in A > B. 2.7 : Fuzzyfications and Defuzzifications Q.19 Write short note on fuzzification. Ans.: ¢ Fuzzification is the process of making a crisp quanti fuzzy. In the real world, hardware such as a digital voltmeter generates crisp data, but these data are subject to experimentaj error. Inputs ee Outputs Fuzzification Inference Defuzzification Fig. Q.19.1 Fuzzification * For each input and output vari more membership functions (Mi more. Fuzzification is the proce: into a fuzzy value. This is achi fuzzifiers (membership functions), lable selected, we define two or F), normally three but can beuting & Optimization Algorithm, off comp! 2-29 Sets and Logic gpiem which deals with of; mr ginall p and converts it into 4 roblem @ fuzzy logic nat 5 fuzzy i y logic nature of BIC Problem e fuzzification PTOCESS Consists of two basic steps ‘al of e, Nctions, puring the first step the intery, into trapezoidal membership fy , puilding a fuzzy knowledge b, t is Defuzzification a3? Mention. ? Explain different methods of pe ‘ach concept is analyzed ase zificatio ¢ Defuz: ion means the fuzzy-to-crisp conversions » Defuzzification has the capability to reduce : a fuzzy set to a crisp single-valued quantity or as a set. « Defuzzification is the process by which a fuzzy consequent is reduced to a singleton or crisp scalar value in order to provide an interface to a typically scalar ‘action’. Actions are typically associated with a scalar defining a specific measurable property. «The task of defuzzification is to find one single crisp value that summarizes the fuzzy set that enters it from the inference block. Examples might include speed of movement, volume of voice, (angular) direction, or a decrease/increase in health/energy. « Defuzzification operates on the implied fuzzy sets produced by the inference mechanism and combines their effects to provide the "most certain" controller output. * Defuzzification methods are as follows : 1. Maxima method. 2. Centroid method 3. Weighted average method 4. Middle-of-maxima method 5. First-of-maxima or last-of-maxima @ Less than PHOTOCOPY Price ne eee ri‘ e.....t—s—S“SSoft Computing & Optimization Algorithms 2-30 Fuzzy Sets we Lon, Max Membership Principle * For peaked output function, max 7 membership is used. It is also called as height method. Fig. Q.20.1. shows — maximum —_ membership 2 principle, Fig. Q.20.1 * The method is expressed as : Me (Z*) 2 we (2) Where Z* is the defuzzified value. Centroid Method In the centroid method, the crisp value of computed by finding the variable value of the membership centroid method. H(z) the output variable i the centre of Bravity ¢ function for the fuzzy value. Fig. Q.20.2 show, z Zz Fig. Q.20.2 Centroid method ¢ This most familiar defuzzification scheme. The composite outpu: fuzzy set is built by taking the union of all clipped or scalei output fuzzy sets. * The crisp output value is then obtained b: or centre of mass of the shape output fuzzy set. y deriving the centroit represented by the composit @ Less than PHOTOCOPY Priceng & Optimization Algorithms 2.31 Ft pee Mty Seta and Logie ’ ordinate of the centre of mass is the cor require value. Because of the variety of shapes hae . _ 7 tered, the method is computationally intensive. , Jite @).2d2 ane Jue @dz where J denotes an algebraic integration. aighted Average Method » Weighted average defuzzification instead of using a composite output membership function, uses the individual clipped or scaled output fuzzy sets. are «The method takes the peak value of each clipped/scaled output fuzzy set and builds the weighted (with Tespect to the peak heights) sum of these peak values. The output is given by : : Luc @z “Yue @ gr = 2(0.8)+0(0.6) 0.8+ 0.6 0.8a+ 0.6a 14 ©The weighted .average method is formed by weighting each membership function in the output by its respective maximum membership value. z awe. Fuzzy Sets ai Soft Computing & Optimization Algorithms 2 - 32 2 nd Lon | se} symmetrical 9, eThis method can be used only for sy tp, membership functions. Other Methods a. Centre of sums * As with the centroid method, the clipped or scaled output fury, } sets are combined to form a composite output fuzzy set, 7, contribution of each clipped or scaled fuzzy set, however, i considered individually. * Unlike the Centroid method which builds the output compo: set using the union operation the centre of sums method taj the SUM of the clipped/scaled output Fuzzy sets and tt computes the centre of mass of the resulting shape. * Thus overlapping areas, by this technique. b. Singleton method if they exist are reflected more than ong * The singleton method is quite a simple method that is widel used in hardware implementations. The technique uses outpu: membership functions which are just vertical lines of height 1 in output space, or singletons. is a weighted average’ of all qualified output sets, = @ Loss than PHOTOCOPY PriceUNIT - in a | 3 Fuzzy Systems 3. 1 Fuzzy Controller Q1 Explain architecture of fr fenegy# Expnin sep a an cn mm: What rae Gee em. 8. : © Fig. Q. i Ans. : ¢ Fig. Q.1.1 shows architecture of a fuzzy control system. Controlled | process Fig. Q.1.1 Architecture of a fuzzy control up into four main parts : *A typical fuzzy system can be split inference engine and a Fuzzifier, a knowledge base, an defuzzifier. © The fuzzy controller has four main components : 1. Rule base : It holds the knowledge with the set of rules. It helps for controlling the system. a-NFuzzy Systems Soft Computing & Optimization Algorithm: control rules are 2. Inference mechanism : Evaluates which 1" nich type of ich relevant at the current time and then deci input given to the plant. 3. The fuzzification interface : It simply that they can be interpreted and compa! rule-base system. ions reached by 4. Defuzzification interface converts the conclusion: ° lant. the inference mechanism into the inputs to the Pp’ Rule base strategy : trol rules . fuzzy con * Rule base strategy has to do with the incorporated into themselves. The property of completeness ne and engineering fuzzy control rules through design experier knowledge. dition is not An additional rule is added whenever 4 fuzzy et era included in the rule base or whenever the degree oe aie between some inputs and predefined fuzzy com me * than some level. ifies the inputs so moat the rules in the ° Fig. Q.1.2 shows pure fuzzy system. Fuzzy rule base Fuzzy sets in U Fuzzy sets in V Fuzzy inference engine Fig. Q.1.2 Basic configuration of Pure fuzzy systems | | @ Less than PHOTOCOPY Pricesoft Computing & Optimization Algorithine 3-3 steps In fuzzy control system ; Fuzzy Systems te . A FC operates by repeating the cycle : 1 Measurements are taken of all rele ‘uzzification : Con i a mE version of the measur i to express their uncertainties iments into faazy sets ‘vant variables 3, Fuzzified measurements are used by evaluate the coni or more fuzzy eee parece) of this evaluation is one —— ‘universe of possi a 4. Reena : The outcome is co: eee value. y the inference engine to verted into a single crisp eln order to Process the input to get the output reasoning there are six steps involved in the creation of a rule based fuzzy system : 1. Identify the inputs and their ranges and name them. 2. Identify the outputs and their ranges and name them. 3. Create the degree of fuzzy membership function for each input and output. 4. Construct the rule base that the system will operate under 5. Decide how the action will be executed by assigning strengths to the rules 6. Combine the rules and defuzzify the output. To design the fuzzy controller, the control engineer must gather information on how the artificial decision maker should act in the closed-loop system. Q2 Explain open loop and closed loop ¢ Ans.:¢ There are two kinds of control systems : Open-loop control systems and closed-loop control systems. ontrol system. 4. Open loop control system : * Fig. Q.2.1 shows open loop control. © Systems in which the output has no é! are called open-loop control systems. @ Less than PHOTOCOPY Pri {fect on the control actionoa, Fuzzy Systems Fig. 0.2.1 Open loop control there is NO feedback path buy e In an open-loop control system the output. i di only forward path between the input an izzy system us # When used as an opencloop controller eens ie — sets up some control parameters and ¢ es | according to these control parameters. ; + Open-loop system is easier to build and lower in cost. ©The open-loop control system can only be used when the js low and the disturbance can ty requirement for accuracy is ashing machines, ste neglected, such as the control of Ww: 4 Stepper motors. Many applications of fuzzy systems in consumer electronics belong to this category. 2. Closed loop control system : «The control system in which the output has an effect on the control action is called as closed-loop control system. When usec as a closed-loop controller, the fuzzy system measures the outputs of the process and takes control actions on the proces continuously. Applications of fuzzy systems in industh: processes belong to this category. © Fig. Q.2.2 shows close loop control system. Fuzzy system oe : Ig. 0.2.2 Close loop control system © @econes @ Lace the. mise_eBy the use of feedback, soft Computing & Optimization Algurithy, a eas Fuzzy Systems A system that compares the uses the difference as a me control system or feedback Output and the reference input and Ns of control is called a closed-loop control system, ea fopern the closed-loop control system is relatively insensitive to external disturbances and internal variations in system parameters, e The closed-loop control system is generally high in cost and complexity. Q.3 Explain fuzzy control system design problem. Ans. : Control System Design Problem Fuzzy control system design essentially amounts to 1. Choosing the fuzzy controller inputs and outputs, 2. Choosing the Preprocessing that is needed for the controller inputs and possibly post Processing that is needed for the outputs, 3. Designing each of the four components of the fuzzy controller. ® Since fuzzy control is a relatively new technology, it is often quite important to determine what value it has relative to conventional methods. + Fig. Q.3.1 shows basic control system. Process is the object to be controlled. Its inputs are u(t), outputs are y(t) and reference input is r(t). BNE Process (P) FE==*> Output Fig. Q.3.1 Control system @ Less than PHOTOCOPY Price-6 Fuzzy Sef Computing & Optimization lgortone 976 _F¥5 Sy, * Consider the example of cruise control problem : @. u(t) is the throttle input b. y(t) is the speed of the vehicle ¢. r(t) is the desired speed that is specified by the driver, © The plant is the vehicle itself. The controller is the compute; in the vehicle that actuates the throttle based on the speed of the vehicle and the desired speed that was specified. * Fuzzy control design methodology can be used to construct controllers for challenging real world application. Feedbac, control system design problem is defined by using Ronlines Vector valued function h( ). The h( ) is defined with ting reference, u(t) = hit, x(t), r(t)] where u(t) = Process control input r(t) = System reference input x(t) = System state vector * In the process of designing a linear or nonlinear control system, is usually necessary to obtain a number of parameters of tle controller in order to define a good design that meets a numbe of performance requirements under certain practical constraints. In control system desi; determined by the control scheme or c engineer opts to apply. © The feedback control law control system and gives ign, the structure of a controller is usually ‘ontrol law that the desig (h) is supposed to stabilize the feedbat 800d performance, © Full state feedback or output feedback is e ut) = h [x(t] Xpressed as : ut) = Alyy) yat] © Here y() is the system output or response function. @ Less than PHOTOcopy Pricesoft Computing & Optimization Algorithms 3-7 Fuzzy Systems «Function “h” is also applied to single input with single output system. It is expressed as follows : u(t) = e(t) K, «Fuzzy controller design consists of turning intuitions, and any other information about how to control a system, into set of rules. These rules can then be applied to the system. If the rules adequately control the system, the design work is done. If the rules are inadequate, the way they fail provides information to change the rules. Fuzzy Rule Base and Approximate Reasoning Q.4 What is fuzzy reasoning ? Discuss in detail the fuzzy reasoning for single rule with multiple premises and multiple rules with multiple premises. Ans.: Fuzzy reasoning the goal of fuzzy logic is to provide the basis for reasoning under non-binary information. The ensuing reasoning system often this is referred to as “Approximate reasoning" or "Fuzzy reasoning". Fuzzy reasoning might be considered more exact precisely because it does not assume a binary universe. Approximate Reasoning The basis for formal reasoning is an inference procedure, itself based upon an appropriate model for ‘if-then’ rules, or modus ponens. The general goal is to infer the associated with a proposition, B, from the implication, A,orA—B. "Consider, '‘A' denotes “sharp corner" and 'B' "approach slowly" then we can naturally express the implication by, | xisA | IFxis A, THEN y is B premise 1 (fact) premise 2 (fact) i consequence (conclusion) | yisB Less than PHOTOCOPY PriceSoft Computing & Optimization Algorithms 3-8 Fuzzy Sei * However, we also take for granted that tere is an img, Monotonic relationship between the ee Satisfaction i. Premise, Thus, a "somewhat sharp" comer enables us to Bent inferring that the approach should be "more or less" slow, op Premise 1 (fact) Pee A Premise 2 (fact) : IF xis A, THEN Yisg —_ consequence (conclusion) : y is B Definition fuzzy reasoning : Let A and A’ be fuzzy sets Universe X and B a fuzzy set on Y. Implication, A B, in terms of a fuzzy relation R on the Cartesian Product, X xy hb fuzzy proposition B’ induced by the premise "x is A" and the fu rule “if x is A then y is B" is defined in the form of the fy composition, on , iS des, . Bo = A’o R=A’o (A B) © IE we go on to assume the max-min operator, then we general model of fuzzy reaso the specific case of composition based ¢ have the special case of the aby ning, Hey) = V (dadx) a BR y) xea! “max-min [( (x), (x, yl xe A min Fig. 0.4.4 Single rule ° Wearenow in a Consider the follow; with single Premise Position to build Some reasoning &S ‘ing three Special cases, @ Less than PHOTOcopy Pricegoncomyuting & Optimization Algoritoms 3.9 Fuzzy Systems 1, single rule with single premise : The Premise simplifies to the special case of a scalar thresholds, Hp = WA paly) >, single rule with multiple Premises . the special case of the minimum of the compositional rule implies : The premise simplifies to two scalar thresholds. Thus, C = (AB) © (AxB4 © or HBo= win W2 Altc(z) min Fig. Q.4.2 Single rule with multiple premises 3. Multiple rules with multiple premises : The ‘max’ operator of equation now applies, thus the area of the implication is the maximum of each minimally thresholded premise. | | min ————— Fig. Q.4.3 Multiple rules with multiple premises @kcove se Less than PHOTOCOPY Price-10 Fur: Soft Computing & Optimization Algoritlnns 2 Y Sieg ng is divided into four steps : 2. Firing strength 4, Overall output MF © Process of fuzzy reason! 1. Degrees of compatiability 3. Qualified consequent MF Advantages of fuzzy system 1. Robust approach to solve many i lex systems, 2. Employable in very comp! " mathematical model for highly nonlinear processes. 3. Hence, low computational costs and ease at using jt a embedded systems. 4. Expert knowledge in complex systems can be formulated ; ordinary language. Disadvantages 1. The number of rules can grow exponentially inverse with t, | accuracy level. Undesirable high complexity and rule-chaini; | problem (Castro, 1999). 2. The rules and the membership function for (imprecise) d: must be (accurately) known and defined. real-world problems. when there is no sin, te 3. Must be combined with an adaptive system (such as nevi networks) if some heuristics is desired. 3.3 Decomposition of Compound Rules and Aggregation of Fuzzy Rules Q.5 Write short note on fuzzy propositions Ans.:¢ In propositional logic the j : . tules are just a form Proposition and they Must be true or false. In fact any propositt may be combined with other propositions in a process whith! called a logical reasoning, a ¢ In case of fuzzy Proposition th values it varies from 2 to n, fast, medium, e Tange is not confined to only ™