International Research Journal of Engineering and Technology ( I RJET) e-ISSN: 2395 -0056
Volume: 02 Issue: 04 | July-2015 www.irjet.net p-ISSN: 2395-0072
© 2015, IRJET.NET- All Rights Reserved Page 940
A HYBRI D FI REFLY BASED APPROACH FOR DATA CLUSTERI NG
Gunjan Dashora
1
, Payal Awwal
2
1
Student, Computer Science, Govt. Women Engineering College,Ajmer, Rajasthan, India
2
Assistant Professor,Computer Science, Govt.Women Engineering College, Ajmer ,Rajasthan,India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In data mining, clustering or cluster
analysis is a common technique. Clusters is a set of
objects which are assigned into a group. The objects in
a single cluster are similar to each other but they are
different from the objects in other clusters. A collective
behavior of social systems like insects such as ants, fish
schooling ,honey bees and birds are called as swarm
intelligence(SI).In this paper, a swarm intelligence
based technique for data clustering is proposed using
firefly and nelder mead search method .K-means
algorithm tends to conver ge faster than firefly
algorithm but usually trapped in a local optimal area. A
new way of integrating firefly with nelder mead search
method proposed in this paper.
Key Wor ds: Firefly algorithm, nelder mead simplex
search, swarm intelligence, K-Means.
1. INTRODUCTION
The process of grouping of data into number of clusters
are known as data clustering. The aim of data clustering is
to keep the objects in a single cluster are similar to each
other but they are different from objects in other
clusters.[1]. K-means is the most popular and widely used
method for clustering.
In k-means , we use Euclidean distance for good clustering
results.[2].However k-means contains several drawbacks
like trapped into local optima and local minima and it is
also sensitive to initial cluster centers.[3][4].
A new prototype Swarm Intelligence(SI) is being used in
research settings to improve the management and control
of large numbers of collaborating entities like computer
and sensor networks, communication, satellite
constellations and many more. It is a collective behavior of
insects.[5] Firefly algorithm is used to overcome the
problem of local optima in k-means algorithm. The firefly
algorithm works on the flashing behaviour of firefly, but
the convergence rate of Firefly algorithm is slower than
those of local search technique( Nelder Mead simplex
search). To deal with the the slow convergence of firefly
,we combine firefly with nelder- mead simplex search.
2. K-MEANS ALGORITHM
Semi structured or unstructured datasets are classify with
the help of k-means clustering. K-means clustering is
simple and it has the ability to handle voluminous
datasets. Therefore, this is the one of the most common
and effective method to classify data.
The parameter used in k-mean clustering is the number of
clusters and the initial set of centroids. The distance of
each item in the dataset is calculated with each of the
centroids of the respective cluster. The item is then
assigned to the cluster with which the distance of the item
is least. The centroid of the cluster to which the item was
assigned is recalculated.
The standard k- means algorithm is as follows-
Initial positions of K cluster centers are determined
randomly. Following phases are repeated:
a)For each data vector : the vector is allocated to a cluster
which its Euclidean distance from its center is less than
the other cluster centers. The distance to cluster center is
calculated by Eq. (1):
Dis( Xp. Zj ) = pi - Zji )
2
(1)
In Eq (1), Xp is p
th
data vector, Zj is j
th
cluster center and D
is the dimension of data and cluster center.
b) Cluster center are updated by Eq (2):
Zj = [ ] (2)
In Eq.(2), nj is the number of data vectors corresponding
to j
th
cluster and C j is a subset of the total data vectors
which constitute j
th
cluster and are in it.
Phases (a) and (b) are repeated until stop criterion is
satisfied.
3. FI REFLY ALGORI THM
Short and rhythmic flashes for communication and
attracting the potential hunt are used by most of the
fireflies. Yang introduce this firefly algorithm in 2008[6].
Firefly works on three rhapsodize rules-
1)All fireflies are unisex, so that one firefly will be
attracted to other fireflies regardless of their sex.
2)Attractiveness is proportional to their brightness. Thus,
for any two flashing fireflies, the less brighter one will
move towards the brighter one. The attractiveness is
proportional to the brightness and they both decrease as
their distance increases. If there is no brighter one than a
particular firefly, it will move randomly.