A Holistic Review on Gravitational Search Algorithm and its Hybridization with other Optimization Algorithms

Presented by
Sajad Ahmad Rather
Full Time Research Scholar
Department of Computer Science
School of Engineering and Technology
Pondicherry University
122-02-2019

 Objective
 Gravitational Search Algorithm (GSA)
 Advantages and Disadvantages of GSA
 GSA Hybridization
 Discussion
 Conclusion and Future Scope
 References
222-02-2019

 A Gravitational search algorithm is a physics-based
heuristic algorithm inspired by Newton’s gravity law. GSA is
good at finding the global optimum but has the drawbacks
of slow convergence speed and getting stuck in local
minima in last iterations.
 To overcome these problems, the GSA is hybridized with
other swarm based optimization algorithms and it results in
the increase in searching capability, problem-solving and
application domains of the gravitational search algorithm.
322-02-2019

 The GSA has been used to solve various optimization
problems in different application areas such as
clustering, classification, feature subset selection, load
power dispatch, routing, etc. and it shows better
performance than other swarm intelligence algorithms.
 This paper gives information about the GSA and its
hybridization with other meta-heuristic algorithms.
422-02-2019

 The Gravitational Search Algorithm (GSA) is a powerful
heuristic optimization method, which is based on the
concept of mass interaction i.e. “a particle in the universe
attracts every other particle with a force that is directly
proportional to the product of their masses and inversely
proportional to the square of the distance between
them”.
 If and are two point masses, R is the distance
between them, then the gravitational force, F is
calculated by using equation as under:
1 2
2
M M
F G
R

1M 2M
522-02-2019

Advantages of GSA
 Simple implementation.
 Suitable swarm-based algorithm for solving non-
linear optimization problems.
 It takes less computational time.
 Generation of feasible solutions.
 Adaptive learning capability.
 Gives high precision results.
722-02-2019

Disadvantages of GSA
 The GSA operators are complex in nature.
 The searching capability of GSA gets slower in the
last iterations of the algorithm.
 It is not flexible due to the inactivity after
convergence.
 Randomized nature of Gravitational operator (G) in
the algorithm.
822-02-2019

 Hybridization is the technique of modifying the
mathematical structure of the parent algorithm by
using another optimization algorithm(s) so that the
limitations of the parent algorithm can be
removed.
 The main advantage of hybridization is to increase
the search space and the problem-solving domain
of the algorithm.
 It also improves the exploration and exploitation
capabilities of the algorithm.
922-02-2019

Scope of GSA Hybridization
1022-02-2019

Clustering
 Clustering is the technique of grouping similar data samples based
on distance and similarity criterion of the data points.
 The K-harmonic clustering is the most commonly used clustering
algorithm because it has simple implementation and takes fewer
iterations. But there is a major drawback in K-harmonic algorithm i.e.
its dependency on initial states of the data centers.
 Here, GSA helps the clustering algorithm (i.e. KH means) to get away
from “trapping local optima” problem and also increases its
convergence speed.
1122-02-2019

Classification
 Classification is the process of categorizing the data into groups
based on mathematical information. It is basically a technique of
finding the patterns in data and is a pattern recognition method.
 The GSA is combined with K-nearest neighbor for the classification
of data. The GSA provides the randomized initialization of the search
space and increases the optimization of the features. The hybrid
technique is tested on 12 benchmark datasets.
 Prototype generation is the process of reducing the dimension of the
class samples used for decision making. The GSA is hybridized with
K- nearest neighbor for classification of prototypes.
1222-02-2019

Feature Selection
 Feature selection is one of the fundamental steps in the
classification process in data mining. It simply means to
get the important features from the data set that can
reduce the dimensionality and search space of the
problem.
 It is carried out using GSA, a swarm based technique and
Optimum-Path Forest (OPF), a powerful pattern
recognition method. The GSA acts as the optimization
tool that helps in finding the pattern(s) in the search
domain and maximizes the output given by OPF.
1322-02-2019

Neural Networks
 The GSA is used with a genetic algorithm for training the
neural network. It is used for performing the global
search in order to find the global optimum and then
genetic algorithm is used for performing the local search
around the solution. The hybrid algorithm shows better
performance than the back propagation algorithm.
 GSA is hybridized with PSO for training the multilayer
FNN. The hybridization results in good convergence
speed and avoidance from the “trapping in local optima”
problem for GSA.
1422-02-2019

Power Systems
 GSA and PSO are used to solve the load dispatch
problem in power systems by considering some
constraints such as Generator rate, transmission
loss, etc. The proposed algorithm shows better
performance than other power system optimization
algorithms.
1522-02-2019

Routing
 The GSA is a memory less algorithm i.e. it’s searching
operator considers only the current position of the
agents. This problem can be solved by using PSO which is
a memory based algorithm.
 To increase the quality of the optimal solutions, the
social operators of the PSO are combined with GSA
operators.
 The improved GSA-PSO hybrid algorithm is used for path
planning which is a global optimization problem.
1622-02-2019

Optimization
 Optimization is the technique of selecting the most
feasible solution for the given problem. The GSA is also
hybridized with other algorithms for solving the
optimization problems in different fields.
 The artificial immune system (AIS) algorithm is hybridized
with GSA in order to overcome the drawback of local
minima trapping in GSA.
 Gravitational operators are used for increasing the fitness
value of the particles in the PSO algorithm. The hybrid
PSO-GSA hybrid algorithm results in the decrease in
computational cost and increase in the feasibility and
efficiency of the PSO.
1722-02-2019

GSA Hybridization Applications
1822-02-2019

Year wise analysis of GSA Hybridization
1922-02-2019

 This paper provided a comprehensive survey of GSA and
its hybridization with other optimization techniques.
Now, it can be concluded that GSA is a powerful swarm
based optimization technique that has profoundly
impacted the various application areas of different fields
of study whether it be clustering, classification, feature
extraction, Routing, and Neural Networks in computer
science or Emission load power dispatch and power
distribution in power systems.
 GSA is an ideal population-based algorithm suitable for
solving different optimization problems that were
previously tackled using classical optimization techniques
such as GA and PSO; with higher accuracy and efficiency.
2022-02-2019

 This fact can never be underestimated that GSA has a lot
to offer preferably to the computer science community as
it can be hybridized with other heuristic algorithms such
as Biogeography-based optimization, Bacterial Foraging
Algorithm, and Grey Wolf Algorithm because these
optimization techniques can be used for overcoming the
drawbacks such as global optimization and premature
convergence of GSA.
 In recent years, it can be observed that Big Data, IoT,
Cloud Computing and Information Security are hot areas
of computer science and quite popular in the research
community. The optimization capability of GSA can be
utilized in these areas.
2122-02-2019

[1] E. Rashedi, H. Nezamabadi-Pour and S. Saryazdi, “GSA: A Gravitational Search
Algorithm, ”Information Sciences, 179(13), 2232–2248. 2009.
[2] E. Rashedi, H. Nezamabadi-Pour and S. Saryazdi, “BGSA: binary gravitational search
algorithm,” Nat. Comput., vol. 9 , pp. 727-745, 2010.
[3] David Halliday, Robert Resnick and Jearl Walker (Extended) , Fundamentals of
Physics, 6th Edition, Wiley, 2000.
[4] Sun and Zhang, “A hybrid genetic algorithm and gravitational using multilevel
thresholding,” Pattern Recognit. Image Anal., pp. 707–714, 2013.
[5] Minghao, Hu, Yang, and Li,“A novel hybrid K-harmonic means and gravitational
search algorithm approach for clustering,” Expert Syst. Appl., vol. 38 , pp. 9319-9324,
2011.
[6] Han, et al.,“Feature subset selection by gravitational search algorithm optimization,”
Inf. Sci., vol. 281, pp. 128-146, 2014.
[7] Hatamlou, Abdullah, H. Nezamabadi-pour, and Nezamabadi, “A combined approach for
clustering based on K-means and gravitational search algorithms,”Swarm Evol.
Comput., vol. 6, pp. 47–55, 2012.
2222-02-2019

[8] Ghalambaz, et al., “A hybrid neural network and gravitational search
algorithm method to solve well known wessinger’s
equation,”International journal of MAIMM Engineering, vol. 5, 2011.
[9] Xiang,et al.,“A novel hybrid system for feature selection based on
an improved gravitational search algorithm and k-NN method,” Appl.
Soft Comput., vol. 31, pp. 293-307, 2015.
[10] Yu, Wu, Ying, and wang.” Immune gravitational inspired
optimization algorithm. In International Conference in Intelligent
Computing,” Spinger, 2011.
[11] J. Papa, et al., “ Feature selection through gravitational search
algorithm,” Acoustics Speech and Signal Processing (ICASSP), IEEE
International Conference, 2011.
[12] Jiang, Ji, and Shen, “A novel hybrid particle swarm optimization
and gravitational search algorithm for solving economic emission load
dispatch problems with various practical constraints,” Int. J. Electr.
Power Energy Syst. vol.55, pp. 628–644, 2014.
2322-02-2019

[13] Xiangtao,Yin, and Ma, “Hybrid differential evolution and
gravitation search algorithm for unconstrained optimization,” Int. J.
Phys. Sci. vol.6, pp. 5961–5981, 2011.
[14] Seyedali Mirjalili and Amir Gandomi, “Chaotic gravitational
constants for the gravitational search algorithm,” Applied Soft
Computing, vol. 53, pp. 407–419, 2017.
[15] Mirjalili, Hashim, Sardroudi and H.M. “Training feedforward
neural networks using hybrid particle swarm optimization and
gravitational search algorithm,” Appl. Math. Comput. vol. 218, pp.
11125–11137, 2012.
[16] Indu Bala and Anupam Yadav, “Gravitational Search Algorithm: A
State-of-the-Art Review,” Harmony Search and Nature Inspired
Optimization Algorithms, Advances in Intelligent Systems and
Computing, vol. 741, pp.27-37, 2018.
[17] Yugal kumar and G. Sahoo, “A Review on Gravitational Search
Algorithm and its Applications to Data Clustering,” I.J. Intelligent
Systems and Applications, vol. 6, pp. 79-93, 2014.
2422-02-2019

[18] J.Kennedy and R. Eberhart, “ Particle swarm optimization, “ in: IEEE
International Conference on Neural Networks, pp. 1942-1948, 1995.
[19] A. Kalinlia and N. Karabogab, “Artificial immune algorithm for IIR filter
design Engineering,”Applications of Artificial Intelligence, vol. 18, pp. 919-
929, 2005.
[20] K.S. Tang, K.F. Man, S. Kwong and Q. He, “Genetic algorithms and their
applications,” IEEE Signal Processing Magazine, vol. 13,pp. 22-37, 1996.
[21] M. Dorigo, V. Maniezzo and A. Colorni, “The ant system: optimization by
a colony of cooperating agents,” IEEE Transactions on Systems, Man, and
Cybernetics – Part B, vol. 26, pp. 29-41, 1996.
[22] Storn and Price , “Differential evolution-a simple and efficient heuristic
for global optimization over continuous space,” J. Global Optim., vol. 11,
pp. 341-359, 1997.
[23] S. Kirkpatrick, C.D. Gelatto and M.P. Vecchi, “Optimization by simulated
annealing Science,” vol. 220, pp. 671-680, 1983.
2522-02-2019

[24] Dan Simon,” Biogeography- Based Optimization,” IEEE Transactions on
evolutionary Computation,vol. 12, No. 6, December 2008.
[25] Chaoshun Li and Jianzhong Zhou, “Parameters identification of hydraulic
turbine governing system using improved gravitational search algorithm,”
Energy Conversion and Management, vol. 52, pp. 374–381, 2011.
[26] M. Soleimanpour-moghadam, H. Nezamabadi-pour and M. M. Farsangi ,
“A quantum behaved gravitational search algorithm,” In: proceeding of Int.
Conf. Computational Intelligence and Software Engineering, Wuhan, China,
2011.
[27] LI Pei and Duan HaiBin, “Path planning of unmanned aerial vehicle
based on improved gravitational search algorithm,” Science China published
by Springer, vol.55, pp. 2712–2719, 2012.
[28] Soroor Sarafrazi and Hossein Nezamabadi-pour, “Facing the
classification of binary problems with a GSA-SVM hybrid system,”
Mathematical and Computer Modeling, vol. 57, pp. 270–278, 2013.
2622-02-2019

[29] Hatamlou , Abdullah, and Othman, “Gravitational Search Algorithm with
Heuristic Search for Clustering Problems,” In proceeding of 3rd IEEE on
Data Mining and Optimization (DMO), pp. 190 – 193, Selangor, Malaysia,
June 2011.
[30] Hatamlou, Pour, and Abdullah, “Application of gravitational search
algorithm on data clustering,” In: published in proceeding of 6th
international workshop on Rough Set and Knowledge technology (RSKT-
11), pp. 337-346,2011.
[31] Y. Jamshidi and V.G. Kaburlasos, “gsaINknn: a GSA optimized, lattice
computing knn classifier,” Eng. Appl. Artif. Intell., vol. 35, pp. 277-285,
2014.
[32] M. Rezaei and H. Nezamabadi-pour, “A prototype optimization method
for nearest neighbor classification by gravitational search
algorithm,”Intelligent Systems (ICIS), 2014 .
[33] M. Rezaei and H. Nezamabadi-pour, Using gravitational search algorithm
in prototype generation for nearest neighbor classification,
Neurocomputing, vol. 157, pp. 256-263, 2015.
2722-02-2019

[34] A. Ghaemi, et al., “Automatic channel selection in EEG signals for
classification of left or right hand movement in Brain Computer Interfaces
using improved binary gravitation search algorithm,” Biomed. Signal
Process Contr., vol. 33, pp. 109-118, 2015.
[35] C. Li and J. Zhou, “Semi-supervised weighted kernel clustering based on
gravitational search for fault diagnosis,” ISA Trans., vol. 53, pp. 1534-1543,
2014.
[36] C.C.O. Ramos, et al., “New insights on nontechnical losses
characterization through evolutionary-based feature selection,” IEEE Trans.
Power Deliv., vol. 27, pp. 140-146, 2012.
[37] J. Xiang, etal., “ A novel hybrid system for feature selection based on an
improved gravitational search algorithm and k-NN method,” Appl. Soft
Comput., vol. 31, pp. 293-307, 2015.
[38] X. Han, et al., “Feature subset selection by gravitational search algorithm
optimization,”Inf. Sci.,vol. 281, pp. 128-146, 2014.
2822-02-2019

[39] F. Barani, M. Mirhosseini and H. Nezamabadi-pour, “Application of
binary quantum-inspired gravitational search algorithm in feature subset
selection,” Appl. Intell., vol. 47, pp. 304-318, 2017.
[40] Rajagopal , Anamika,Vinod and Niranjan,“gsa-fapso-based generators
active power scheduling for transmission congestion management, ” pp. 1 –
8, 2018.
[41] Shanhe,Chaolong, Wenjin and Yanmei, “An Improved hybrid particle
swarm optimzation with dependent coefficients for global optimization,”
33rd Youth Academic Annual Conference of Chinese Association of
Automation (YAC), pp. 666-672, 2018.
[42] Y. Ho and D. Pepyne, “Simple explanation of the no-free-lunch theorem
and its implications,” J. Opt. Theory Appl., vol. 155, pp. 549–570, 2002.
[43] Lavika,Sunita,Sharthak and Satyajit, “Hybridization of gravitational
search algorithm and biogeography based optimization and its application
on grid scheduling problem,” Ninth International Conference on
Contemporary Computing (IC3), pp. 1-6, 2016.
2922-02-2019

SUGGESTIONS
&
QUERIES
3022-02-2019

A Holistic Review on Gravitational Search Algorithm and its Hybridization with other Optimization Algorithms

More Related Content

What's hot (20)

Similar to A Holistic Review on Gravitational Search Algorithm and its Hybridization with other Optimization Algorithms (20)

Recently uploaded (20)

A Holistic Review on Gravitational Search Algorithm and its Hybridization with other Optimization Algorithms