Proceedings of the 11th INDIACom; INDIACom-2017; IEEE Conference ID: 40353
2017 4 International Conference on “Computing for Sustainable Global Development”, 01st - 03rd March, 2017
Bharati Vidyapeeth's Institute of Computer Applications and Management (BVICAM), New Delhi (INDIA)
th
Optimization Control Techniques using Soft
Computing Approaches
Yusuf Perwej
Nikhat Akhtar
Jai Pratap Dixit
Department of Computer Information
System
AI BAHA University, AL BAHA
Kingdom of Saudi Arabia (KSA)
Deptt. of Computer Science &
Engineering,
Babu Banarasi Das University,
Lucknow, INDIA
Deptt. of Computer Science &
Engineering,
Ambalika Institute Of Management and
Technology , Lucknow , INDIA
Email ID:yusufperwej@gmail.com
Email ID: dr.nikhatakhtar@gmail.com
Email ID: jpdixit.iiita@gmail.com
Abstract – Soft computing approaches have different
capabilities in error optimization for controlling the complex
system parameters, Soft computing approaches provide a
learning and desion making support from the relevant datasets or
others experts review experiences. Soft computing optimization
approaches can be variety of many environmental and
alsostability related uncertainties. This paper explain the
different soft computing approaches viz., Genetic algorithms,
Fuzzy logics, results of different error optimization control case
studies. Mathematical Models refer to the Conventional error
optimization control , which define dynamic control Conventional
controllers are often inferior to the intelligent controllers, due to
lack in comprehensibility. The results that controllers provide
better control on errors than conventional controllers.
Hybridization of technique such as fuzzy logic with genetic
algorithms etc., provide a better optimization control for the
designing and devepling of intelligent systems.
Keywords - Fuzzy Logic; Evolutionary & Genetic
Algorithms;Neural Networks; Soft Computing Approaches.
I. INTRODUCTION
Optimization control approaches have been defined for
many decadesand recentlyprovid efficient algorithms with
better performance with appropriate optimization algorithms.[4]
Soft computing optimization techniques is a emerging field
of computational intelligent systems which are integrated
approaches and also highly suitable to determine solution for
different problems such as illdefined , complex and difficult to
modeled.[1] Optimization control define the adjusting the many
inputes to determine the min or max output of results based on
problems[3]. Desirable properties of optimization control
algorithms as perspective of machine are in practice of
execution processes[5]. Defining a algorithms for solving large
scale problems we start by algorithms and identifying solution
process which are extensible to adaptivity.
With reference to define the optimization control problems,
need a cost & space function f which depends on u & p. Soft
computing techniques are enrich with the data sets of many
techniqyes which are complementary within the nature of each
one another. All optimization techniques include cobinotorial
and continuous process uses the fuzzy logics concepts, neural
networks, evoluationary and genetic algorithms.[1] A part of
designing smart intelligent system is a crusical task which
cannot be fulfill by the previous traditional approces of
optimizations. Soft computing optimization paradigam is highly
suitable to implement of many IS as they are real life domain
applications. Soft computing optimization control problems is
refers as a foundation element of the computational intelligence
implementations. Many Industries , business portal problems
are easly solved by the soft computing approaches. Soft
computing approaches for optimization also controls the several
problems which are highly complex and deficult to define.
So we can say that there arethe following properties which
narrated as first one In real life tthere many times become
highly complicated as formulate description of model and it’s
happens that different problems don’t require optimization
control problems issues. And second one is as Different control
optimization approaches deal with the real time applications.
II. MAJOR TYPES OF APPLICATIONS SUPPORTED
Soft computing family discussed with the Fuzzy logic
concepts, Evolutionary Computing methods, Neural networks
as the following :
Suppose a model is using a variational statement for a value
of the parameter p∈ Π, then find the solution of function u =
u(p)
where the functional space V & A are dependent problem.
The parameter p related with a function space for simlycity.
Copy Right © INDIACom-2017; ISSN 0973-7529; ISBN 978-93-80544-24-3
Fig. 1. Soft computing family
102
Proceedings of the 11th INDIACom; INDIACom-2017; IEEE Conference ID: 40353
2017 4 International Conference on “Computing for Sustainable Global Development”, 01st - 03rd March, 2017
th
There are the following supported applications such as
• Intelligent Search :
• Optimization
• Evolutionary Modeling
• Machine Learning
• Desission Support System
• Robotics
• Engineering System
• Genetic Algorithms
A. A. Intelligent Search:
It can be happens with using evolutionary methods. GA
(Genetic algorithms) dose’nt require specific knowledge
problem of intelligent search due to string evoluted with qulity
fitness so search execute made with possible through the string.
They are basically structured .
Fig. 4. Basic process of Evolutionary Modelling
D. Machine Learning
It refers to the supervised learning as well as inductive
learning. Neural Network is most important prime contributors
in specific state area.
Fig. 2. Intelligent binary search
B. Optimization
It is the optimization process for dession making which
satisfy all given constraints. Problem satisfied with all optimal
outcomes . Traditional methods or approaches of optimization
include both gradient based as well as direct different search
approaches. GA satisfied the all the requirement providing
optimum value. The global optimization determine the best
process solution as nonlinear desission making models which
are frequently have a number of sub –optimal solution value.
Fig. 5. Supervise machine learning process
Neural network capable providing incremental learning in
intelligent system designing.
E. Decission Support system:
Different types of decision making have been design in
various area deponds upon the specific domains.
Fig. 3. Optimization Cycle
C. Evolutionary Modelling
Basically EM approaches of optimization process
computing which is based on structure of different problems
determine with the help of process optimization evolution. The
process of control evolution of optimization is change the
inherited traits of a problems.
Fig. 6. Dicission Support System
F. Robotics
In this area soft computing hybridization approaches have
been contributed .
Copy Right © INDIACom-2017; ISSN 0973-7529; ISBN 978-93-80544-24-3
103
Optimization Control Techniques using Soft Computing Approaches
III. OPTIMIZATION
Optimization process control determine the various specific
outputs after processing different contol functions technique or
process which depends upon the various specific input or
variable of problem mechanism. The optimization control
technique using softcomputing describe as the figure 3.9
Fig. 7. Robotics arms optimization
G. Engineering System
Application domain in diffeennt disciplines of engineering
problems likes mechanics,aeronotics etc. have been
implementating using different optimization control techniques.
H. Genetic Algorithms:
There are the following programm applications such as
Genetic Programming, Genetic Algorithms and Evalutionary
Programming
1) Genetic Programming: The didderent operation are
applicable to system control programs in various selected soft
problems with reproduction an existing program by caping
with new and creation of new system programs from existing
system program.
2) Genetic Algorithms:GA used in problems of engineering
and scientific controls applications. They also successfully
applicable to all optimization problems which are determined
to solve using conventional approaches such as m/c learning.
3) Evolutionary Programming:They are usefullapproaches
of optimization when other are impossible to implement. It is
also capable to predict a previous symbol based on symbol.
Fig. 9. Optimization Process
As the mechanism perspective of optimization algorithm [1]
summarize as desirable properties such as scalability to large
problems , generatlization and also a specific performance in
the term of execution process. Computer is a perfect tool for
optimization process which influencing the idea can be input in
electronic format. Any Soft computing implementation
technique for optimization control deponds upon mathematical
theories which used to guarantee the performance of
development.
There are thespecific optimization technique classified such
as Combinatorial Optimization , Continuous Optimization,
Branch-and-Bound (BB) for Global Optimization and
Alternating Minimization
A. Combinatorial Optimization:
Combinatorial control optimization discuss the technique in
which the stages of unlabled points may be explicit
optimization variables problems. Many control techniques
describe a sytandard of supervisec as subroutine problems over
the minimization for different fixed values. Control problems
testing related to the many type of combinatorial problems in
soft computing. Optimization control technique tackle with the
classification as two category:[11] Exact Methods,
Heuristic(approximate) Methods-Exact methods provide
solution in hand but not sutable for the real life as necessary
computation due to their complex nature.(not relastic for the
large complex problems.
Heuristic with meta heuristic as practically seek to find high
quality solution (not optimal) with in resanable execution
time.[14] Neural Network(NNS) [3] use in area of information
driven rather than data driven.with good performance It
Include the common serach methodology. Common mata
search methodology performance is very slowin local opyima
so faster coverage speed cover by dalta bar dalta training
algorithm.
Fig. 8. Genetic Algorithms process
There are above applications in different field medicine,
mathematics, social science ,management,and education etc.
B. Continuous Optimization
By Using continuous optimation for fixed argminy which
eliminating continuous function objective which based on
Copy Right © INDIACom-2017; ISSN 0973-7529; ISBN 978-93-80544-24-3
104
Proceedings of the 11th INDIACom; INDIACom-2017; IEEE Conference ID: 40353
2017 4 International Conference on “Computing for Sustainable Global Development”, 01st - 03rd March, 2017
th
cluster assumption.
A. APPROXIMATE MODELLING TECHNIQUE
Soft computing optimization technique effectively solve
real life probems application dealing with uncertainty and
apprimation. It used fuzzy logic , probabilistic reasoning
approaches.
Fuzzy Logic: Fuzzy logic manipulating and representating
data which was not precise . Fuzzy logic gives an interface that
enables approximate capabilities applied to knowledgebased
system. In intelligent system. Such type model enlisted with as
Human Reasoning, Modal of Mathematical, Transaction
(member & non members), Simple Relastively, System
Fluctuation, Descripts or Lingustic
Fig. 10. Optimization Process
C. Branch-and-Bound (BB) for Global Optimization
Fuzzy function optimization mapping with non fuzzy input
values to fuzzy linguistic steps and vice versa for controlling
numeric input data.
Algorithm:
1. Intilise the linguistic variablesn
2.Function Construction s
3. Applied rule base process
4. Conversion fuzzy values
5. Evalutions rule
6. Result combination
7. Convert –output data as non fuzzy values
8. Probalistic Reasoning
Probabilistic Reasoning deals with uncertainty ubiquitous
property of knowledge property . Softcomputing also uses the
theory of probabilistic
reasoning theory. Incomplete
knowledge causes the uncertainty as disagreement between
information sources in approximation with measurement error.
Fig. 11. Global Optimaization
BB technique used for impractical for large different data
sets.
Which effectively gives a good perfiormance on exhaustive
rsearch for pruning large data search of the problem solution
for many observation.
The aim of probabilistic reasoning and logic is to combilne
the probability capacity for handling uncertainty with detective
capacity to exploit structure. It may be shown as analogous
pattern to fuzzy reasoning as place of fuzziness. Bayesian
Network is a powerfull mechanism and representation of
knowledge using reasoning of probabilistic.
B. Neural Networking Modelling
D. Alternating Optimization Technique
Minimization over the function(f) is a standard training –each
unlabeled stages.
The steps as the follow
1. Initilization (Problem)
2. Evaluation
3.Trach Fitness
4.Modification as per requirement
5.update the mechanism
6. Terminate if process met (3,4)
7.Execute
8. Repeat(2-6)
IV. MODELLING TECHNIQUES
• Approximate Optimization Techniques
• Neural Networking Modelling Technique
• Hybrid Optimization Modelling
Fig. 12. General Structure of Artificial Neural Network
Copy Right © INDIACom-2017; ISSN 0973-7529; ISBN 978-93-80544-24-3
105
Optimization Control Techniques using Soft Computing Approaches
Neural Network Optimization Modelling as soft computing
play a important role as as designing of meachine learning
system by simulating optimization control technique system.
Massively parallel processing of distributed system made up of
interconnected neural computing lelements. It have a ability
acquire knowledge of the whole system 16]. Neural Model are
generly use the simple processors(small amount of local
memory) which recognize specific pattern algorithms to
discover large databases[18] represent normal structure for
optimization.
C. Hybrid Optimization Modeling
Hybrid Optimization Modeling
approaches in
softcomputing derived from many domains such as statistic,
logics and physiology etc. The initial development of position
as control independent methods are utilized among different
community of practitions.Hybridization is capable to eliminate
the limitation and generate problem siolution for individual
approaches. Encoding capability methods involve rule base
aggregation , semantics and antecedent aggregation oparators
and de-fuzzification[25]. Hybrid system of neural network
provide a complex nonlinear relationship which appropriately
suitable for predetermined classification of different domais.But
output precision are often limited and admit zero error.
Hybridization with fuzzy logic is more powerfull designing for
intelligent system. The architecture involve the mechanism
which considerd as adaptive fuzzy system as general
framework. There are the following hybrid modeling
techniques as :Evolutionary Fuzzy hybrid system,Neural Fuzzy And
This leads to:
Gradient Methods: An alternative substitute optimal function
which obtain objective function using gradient technique.as:
A. A. Contineuous Process
The Blalancing function become relatively easy to implement
for all algorithms using balancing constraints such as :
Primal Optimation
optimization in soft
linear classification
boundaries require
technique.
also part of contineuous process for
computing. It related with the directly
and variables in process. Non linear
kernel function to implements the
Neural Genetic Fuzzy
V. ALGORITHMS, OBSERVATION AND RESULT
Basically observation required as the follows :Experimental
Setup, Hyperparameters, Multi-class, Objective value,
Balancing constraint etc.
B. B. Contineous Optimization
This type of optimization control problem illustrate how
independent cluster assumption with given standard Function
(f) determine as objective function as non fuzzy system
Optimazation algorithms based on specific Deterministic
and non deterministic control processing of different parameter.
Algorith for Deterministic Annealing. Here we focus on
using alternate and gradient methods for optimization
Alternate Minimization control Algorithm:
Here the procedure proposed by sindwani 2008 over unlabled
functions.Contribution of two loss terms by C , pi parameter.
Fig. 13. The effective optimization output for unlabled point.
Copy Right © INDIACom-2017; ISSN 0973-7529; ISBN 978-93-80544-24-3
106
Proceedings of the 11th INDIACom; INDIACom-2017; IEEE Conference ID: 40353
2017 4 International Conference on “Computing for Sustainable Global Development”, 01st - 03rd March, 2017
th
Soft computing uses the hybridization of approaches to
overcome the errors and optimize the procedure based on
resulting challenges. These procedures provide may algorithms
for development of new approaches for solving optimization
techniques. Optimization control technique algorithms present
different significant implementation . The optimization
problems can be handle by the different control technique
different approaches with their goal bydemonstrate mentions
algorithms with in different types of simulation codes in
designing of different control issues.
Fig. 14. The degree of non fuzzy system optimized by different parameter
processing
VII. ACKNOWLEDGEMENT
I acknowledge my great gratitude and immense to different
researchers for their encouragement, inspiration and insightful
resurch.
VIII. REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
Fig. 15. The piecewise loss optimized function
[7]
VI. CONCLUSION
This paper explain the different optimization control
technique with significance of soft computing approaches such
as fuzzy logic system methods of evolutionary concepts and
also neural network, which are the prime contributions in
developing and designing different intelligent systems.There
are many N intelligent systems , which require various task
performing such as impressions ,intelligent searchs
,optimization, uncertainty, machine learning and different
automatic dession support making systems etc.
Softcomputing optimationapproaches are very efficient to
explain satisfactory desion making
procedure for the
requirement of
designing and developing of intelligent
envirnments. Fuzzy logic concepts , evolutionary technique
process , neural networks and probalistics reasoning methods
have different capabilities in error optimization for controlling
the complex system parameters. There are major constituents
by hybridratzation such as fuzzy with evolutionary, neural with
evolutionary and neural-fuzzy-evolutionary which have
contributed for achiving a successfully design
and
implementation of intelligent system for next generation. This
paper discussed many contribution of soft computing
optimization process in search ,optimizating, controlling,
machine learning, engineering design and many fuzzy,
linguistic modeling etc. The paper has provide major types of
supported applications, optimization methods and approximate
designing with significant reviews algorithms under utilities
and architecture. Indetitification of technique with performance
of model identification provide a generalizationform of soft
computing.
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
Y. Chen, G. Wang, and S. Dong. Learning with progressive
transductive support vector machine. Pattern Recognition Letter,
24(12):1845–1855, 2003.
I.P. Cabrera et al., Fuzzy Logic, soft computing and applications
in Bio-Inspired Systems: Computational and Ambient
Intelligence, vol. 5517, J. Cabestanyet al., Ed. Salamanca, Spain:
Springer Berlin Heidelberg, pp. 236- 244,2009.
K.B.Mankad,“A Genetic-Fuzzy Approach to Measure Multiple
Intelligence”, Ph.D. thesis, Sardar pateluniv.,vallabhvidyanagar,
India,2013.
K.B.Mankad andP.S.Sajja,“A GFA Driven Framework for
Classification of Multiple Intelligence”, in Proc. of WCECS,
2011, pp.469-473.
F. O. Karry and C. D. Silva, Soft computing and intelligent
system design: Theory, tools and applications, first ed., New
York: Pearson,2004.
S. Rajsekaran and GAV. Pai, Neural Networks,Fuzzy Logic,and
Genetic Algorithms Synthesis and Applications,New Delhi: PHI,
2003.
Bart G. Van Bloemen Waanders, Brain R. Carness
“Optimization
under adaptive error control for a contact tank reactor”,I. proc.
Of IJNME, 2002
C. A. Pena-Reyes, “ Co-evolutionary Fuzzy Modeling”, Ph.D.
dissertation, Swiss Federal Inst. Tech., Lausanne, EPFL,
Switzerland, 2002.
E.Cox, Fuzzy Modeling and Genetic Algorithms for Data
Mining and Exploration, The Morgan Kaufmann Series, CA,
2005.
Kristin P. Bennett ,Emilio Parrado-Hernadez, “ The interplay of
optimization and machine learning research , in proc. Journal of
machine learning research , 2006(1265-1281).
C.
Pena-Reyes,”
Co-evolutionary
Fuzzy
Modelling”
[dissertation]. Federal Inst. Tech.:Univ. Lausanne; 2002.
K.B.Mankad,” The Significance of Genetic Algorithms in
Search, Evolution, Optimization and Hybridization: A Short
Review”, International Journal of Computer Science and
Business Informatics, Vol. 9, No. 1, pp. 103-115, 2014.
D. Chapelle and A. Zien. Semi-supervised classification by low
density separation. In Tenth International Workshop on Artificial
Intelligence and Statistics, 2005.
A.Abraham, Recent Advances in Intelligent Paradigms and
Applications, Jain L. and Kacprzyk J. (Eds.), Studies in
Fuzziness and Soft Computing, Springer Verlag Germany, pp.135, 2002.
Copy Right © INDIACom-2017; ISSN 0973-7529; ISBN 978-93-80544-24-3
107
Optimization Control Techniques using Soft Computing Approaches
[19] Ramaprabha R. and Mathur B.L. ,soft computing optimization
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
for solar photovoltaic arrays. In proc. APRN journal of
engineering and applied sciences , vol 6,n0. 10, octo 2011,18196608
P.Bajpai, and M.Kumar, Genetic Algorithm – an Approach to
Solve Global Optimization Problems, Indian J. Computer Sci.
and Eng., vol. 1, no. 3, pp. 199-206, Oct.-Nov. 2010.
K.B.Mankad, and P.S.Sajja, “The Impact of Genetic Fuzzy
Modeling for Machine Intelligence”, Information Technology
Research Journal.Vol .3(1), pp. 1 – 8, May, 2013.
G. Serag-Eldin, S. Souafi-Bensafi, J. K. Lee,W.K.Chan,M.Nikravesh,Web Intelligence: Web-Based BISC
Decision Support System (WBISC-DSS), in: Y.- Q. Zhang, et
al.,(Eds.), Computational Web Intelligence, Univ. of CA,
Berkely,2004, pp. 391-429.
R. Collobert, F. Sinz, J. Weston, and L. Bottou. Large scale
transductive SVMs. Journal of Machine Learning Research,
7:1687–1712, 2006.
Olivier Chapelle , Vikash Sindhwani “ Optimization techniques
for semisupervised support vector machines , in proc. Journal of
machine learning , 2008, 203-233
W.M. Cheung and U. Kaymak, A fuzzy logic based trading
system, in Proc. of the Third European Symposium on
Natureinspired Smart Information Systems, St. Julians, Malta,
2007,pp. 141-148,
T.Hayat andK.M.J. Knanim,“New Fuzzy CSP for an Optimized
Mobile Robot’s Path Tracking using Genetic Algorithms”,
International Journal of Computer and Information Technology
(ISSN: 2279 – 0764), Vol.3 (3), May 2014,pp.523-531.
Rundolf, Christian, “ The role of soft computing in intelligent
data analsysis: in proc. Ijermt . in 2014
O.Cordon, F.Herrera, F.Hoffmann andL.Magdalena, Genetic
Fuzzy Systems Evolutionary tuning and learning of fuzzy
knowledgebases,Singapore: World Scientific, 2001.
Kunjal Bharatkumar Mankad, An architechural perspective of
soft computing methods. In proc. Ijrmt , in 2015,vol-4,issues-2
2278-9359
Bengio and Y. Grandvalet. Semi-supervised learning by entropy
minimization. In Advances in Neural Information Processing
Systems, volume 17, 2004
Copy Right © INDIACom-2017; ISSN 0973-7529; ISBN 978-93-80544-24-3
108