Databases Reference
In-Depth Information
Fig. 9.3.
Fuzzy swarm net model.
is treated as the personal best or pbest of it. Each net uses both gbest and
pbest values to update its weights for the next iteration as follows:
v
k,d
(
t
)=
τ
(
v
k,d
(
t
−
1) +
φ
1
(
p
k,d
−
x
k,d
(
t
−
1)) +
φ
2
(
p
gd
−
x
k,d
(
t
−
1)))
x
k,d
(
t
)=
x
k,d
(
t
−
1) +
v
k,d
(
t
)
where
k
is the swarm size,
d
=
i
j, w
i,j
obtains its value from
x
k,d
.
The stopping criterion may be allowing the nets to iterate till they
converge to a single decision. However in this process, the nets get
over trained leading to poor performance of the classifier. From different
simulations of a dataset, a suitable range of iteration can be fixed for it.
Different dataset require different range of iterations, one range may not
be suitable for all the datasets. The following high-level pseudocode gives
more insight view of the proposed model.
∗
