IJSRD - International Journal for Scientific Research & Development| Vol.

6, Issue 06, 2018 | ISSN (online): 2321-0613

A Comparison Study of Fuzzy Logic, Neural Network & Neuro – Fuzzy

System
Dhivya Devi S.1 Ravanan R.2
1
Research Scholar 2Principal
1
Department of Statistics
1,2
Presidency College, Chennai-05, India
Abstract— Data mining is a used in large databases to find the rules. These approaches as to derive the reasonable
out use full patterns and trends that go beyond analysis. Data conclusion. The three components are named as
mining plays a vital role in many fields. Health care Industry 1) Fuzzification
composed of large technique amount of database which can The linguistic form of the fuzzy rule is expressed in fuzzy sets
be refined to get valuable information. In this paper the author where these variables are defined in terms of degree of their
presents the fundamental concepts and parameterized association membership function. This method of calculating
comparison in the aspects of neural network, Fuzzy logic, and the degree of belongingness of the crisp input into fuzzy set
Neuro-fuzzy system. is known as Fuzzification.
Key words: Fuzzy Logic, Neural Network, Neuro- Fuzzy 2) Fuzzy inference system
System, Fuzzy Inference System  FIS is based on fuzzy rule.
 Membership values are mapped to output through if-then
I. INTRODUCTION rules.
Now a day’s technology plays a strong role in day-to day life. 3) Defuzzification
An Artificial intelligence has the many different approaches Defuzzification is used for the final fuzzy output, when the
that acquire that human intelligence into machines. A lot of crisp value of the system required.
research work has been carried out by neural network and
fuzzy logic. This paper presents the architecture and the basic
learning of neuro-fuzzy inference system with the
comparison of soft computing approaches such as artificial
neural network and fuzzy logic. Artificial neural network and
fuzzy logic are two techniques involved in problem-solving
approaches in terms of accuracy, human thinking, decision
making and classification. The main aim of the study is to
represent the analytical comparison in between these
techniques, whether to perceive input and give the desired
output by processing the data.

II. FUZZY LOGIC

Fig. 1: Structure of Fuzzy Logic System
The concept of fuzzy logic was proposed by Lotfi Zadeh, a
professor at the University of Claifornia at Berkeley in 1965.
III. ARTIFICIAL NEURAL NETWORK
An organized method for dealing with imprecise data is called
fuzzy logic. The data are considered as fuzzy sets. Fuzzy set A Neural Network in the field of an artificial neural network
is a superset of conventional logic and contains similarities is an interconnected collection of neurons that uses a
and differences with Boolean logic. Fuzzy logic membership mathematical computation for information processing. An
functions are restricted to 0 and 1. Fuzzy set theory generalize artificial neural network is neutrally implemented
classical set theory in which the membership degree is not mathematical model. There exist a large number of highly
restricted to 0 to 1, but the value in [0,1]. By expanding the interconnections processing element called neurons in ANN.
idea of fuzzy sets and fuzzy relations we can define fuzzy The interconnections with their weighted linkages hold the
logic systems (FLS). Fuzzy Logic Systems are one of the informative knowledge. The input signals arrive at the
main developments and successes of fuzzy sets and fuzzy processing elements through connections and connecting
logic. FLS are rule based systems in which an input is first weights. The processing elements of the ANN have the ability
converted from a classical set to a fuzzy set and the processed to learn, recall and generalize from the given data by
by an inference engine that retrieves knowledge in the form adjustment of weights. The computation power can be
of fuzzy rules contained in a rule base. The fuzzy sets demonstrated only by the collective behaviour of neurons,
computed by the fuzzy inference as the output of each rule. A and it should be noted that no single neuron carries specific
FLS is a rule based system that implements a nonlinear information. The models of ANN are specified by the three
mapping between in inputs and outputs. basic entities namely
The structure of fuzzy inference system made up of 1) The model’s synaptic interconnections
three principle components which are rule base comprising of 2) The training or learning rules adopted for updating and
the selected fuzzy rules, a database defining the membership adjusting the connecting weights
functions of the fuzzy rules, and a reasoning mechanism 3) Their activation function.
which performs a fuzzy reasoning inference with respect to The Main advantage of Artificial Neural Network

All rights reserved by www.ijsrd.com 464

 A Comparison Study of Fuzzy Logic, Neural Network & Neuro – Fuzzy System
(IJSRD/Vol. 6/Issue 06/2018/109)

1) Adaptive Learning IV. NEURO-FUZZY SYSTEM

An ANN is endowed with the ability to learn how to do The Neuro-fuzzy systems are soft computing methods of
tasks based on the data given for training. design for intelligent systems that combine various ways of
2) Self – Organization neural network and Fuzzy logic concepts. The combination
An ANN can create own organization of the information of these techniques leads to development of integrated
it receives during learning or learning. system. The neuro-fuzzy system brings the low level learning
3) Real- Time Operation and computational power of neural network into fuzzy system
ANN computations may be carried out in parallel. and the high level learning of fuzzy system into neural
4) Fault Tolerance network. Due to this approach the neural network improves
Network capabilities may be retained even after major their transparency to come closer fuzzy logic system, while
damage. fuzzy system can self-adapt, that make them closer to neural
networks.
In the field of fuzzy modelling research, the Neuro-fuzzy is
divided into two areas:
1) Linguistic fuzzy modelling focused on
interpretability.
2) Precise fuzzy modelling focused on accuracy.

Fig. 2: Simple Neural Network

S.No Parameter Fuzzy logic ANN Neuro-fuzzy system
1. Variable Linguistic variable Weighted number Both
2. Human thinking No Yes Yes
Decision
3. Yes No Yes
making ability
Yes, No, it can classify
4. Accuracy of Results Yes
It gives accurate result result
Yes, It was designed to deal with
5. Uncertainty handling No Yes
uncertainty.
6. Learning ability No Yes Yes
7. Knowledge extraction Yes No Yes
8. Interoperability Yes No Yes
9. Usability of prior knowledge Yes No Yes
10. Problem solving approach Non linear Non linear Linear
Table 1: Parameterize Comparison of Fuzzy Logic, Artificial Neural Network, & Neuro- Fuzzy System
Transactions on Systems, Man and Cybernetics, Part B,
V. CONCLUSION 29(6), 859-870.
The table 1 gives the basic structure for the comparison of [5] Walia.N, Singh.H, & Sharma.A, ―”Adaptive Neuro-
soft computation approaches such as ANN, Fuzzy logic and Fuzzy Inference System- A Survey”, International
Neuro- fuzzy system. The problem solving approaches are Journal of Computer Applications ,ISSN :0975 –
different in terms of accuracy of results, human thinking, 8887),Volume 123 – No.13, August 2015
decision making approaches, etc. NFS is very strong to deal [6] Sonali.m.blind & Priyanka , ―”Research Paper on Basic
with any kind of the problem whether data in uncertain or in of Artificial Neural Network”, International Journal on
complete is to given properly than single fuzzy logic or neural Recent and Innovation Trends in Computing and
network systems. Communication, ISSN: 2321-8169, Volume: 2 Issue: 1,
pg no. 96 – 100
[7] Mousavi J., Ponnambalam K., Karray F., ―”Inferring
REFERENCES
operating rules for reservoir operations using fuzzy
[1] Ajith Abraham, ―”Beyond Integrated Neuro-Fuzzy regression and ANFIS”, Fuzzy Sets and Systems,
Systems: Reviews, Prospects, Perspectives & Elsevier, vol. 158, pp. 1064–1082, 2007.
Directions”, Australia—3842 2. [8] Hadi Sadoghi Yazdi, S.E. Hosseini, Mehri Sadoghi
[2] Arora, N. & Saini, J.R., “A Literature Review on Recent Yazdi, , ―”Neuro-fuzzy based constraint
Advances in Neuro-Fuzzy Applications”, International programming”, Elsevier, Volume 34, Issue 11,
Journal of Advanced Networking Applications (IJANA) November 2010, Pages 3547–3559
, pg no. 14-17,ISSN No. : 0975-0290 [9] Vieira.J, Fernando.M. Dias, Alexander, ―”Neuro-Fuzzy
[3] George j. Klir & Yuan, ―”Fuzzy Sets and Fuzzy Logic: Systems: A Survey “,online publication, researchgate ,
Theory and Applications”, PHI, New Delhi, India 1996. Jan 2015
[4] Quek, C., & Zhou, R. W. (1999). "POPFNN-AAR(S): a
pseudo outer-product based fuzzy neural network." IEEE

All rights reserved by www.ijsrd.com 465