IEEE Access,
Journal Year:
2020,
Volume and Issue:
8, P. 26304 - 26315
Published: Jan. 1, 2020
Multilevel-thresholding
is
an
efficient
method
used
in
image
segmentation.
This
paper
presents
a
hybrid
meta-heuristic
approach
for
multi-level
thresholding
segmentation
by
integrating
both
the
artificial
bee
colony
(ABC)
algorithm
and
sine-cosine
(SCA).
The
proposed
algorithm,
called
ABCSCA,
applied
to
segment
images
it
utilizes
Otsu's
function
as
objective
function.
ABCSCA
uses
ABC
optimize
threshold
reduce
search
region.
Thereafter,
SCA
output
of
determine
global
optimal
solution,
which
represents
values.
To
evaluate
performance
set
experimental
series
performed
using
nineteen
images.
In
first
series,
assessed
at
low
levels
compared
with
traditional
methods.
Moreover,
second
aims
high
six
algorithms
addition
ABC.
Besides,
evaluated
fuzzy
entropy.
results
demonstrate
effectiveness
showed
that
outperforms
other
terms
measures,
such
Peak
Signal-to-Noise
Ratio
(PSNR)
Structural
Similarity
Index
(SSIM).
Alexandria Engineering Journal,
Journal Year:
2023,
Volume and Issue:
68, P. 141 - 180
Published: Jan. 18, 2023
The
use
of
metaheuristics
is
one
the
most
encouraging
methodologies
for
taking
care
real-life
problems.
Bald
eagle
search
(BES)
algorithm
latest
swarm-intelligence
metaheuristic
inspired
by
intelligent
hunting
behavior
bald
eagles.
In
recent
research
works,
BES
has
performed
reasonably
well
over
a
wide
range
application
areas
such
as
chemical
engineering,
environmental
science,
physics
and
astronomy,
structural
modeling,
global
optimization,
engineering
design,
energy
efficiency,
etc.
However,
it
still
lacks
adequate
searching
efficiency
tendency
to
stuck
in
local
optima
which
affects
final
outcome.
This
paper
introduces
modified
(mBES)
that
removes
shortcomings
original
incorporating
three
improvements;
Opposition-based
learning
(OBL),
Chaotic
Local
Search
(CLS),
Transition
&
Pharsor
operators.
OBL
embedded
different
phases
standard
viz.
initial
population,
selecting,
space,
swooping
update
positions
individual
solutions
strengthen
exploration,
CLS
used
enhance
position
best
agent
will
lead
enhancing
all
individuals,
operators
help
provide
sufficient
exploration–exploitation
trade-off.
mBES
initially
evaluated
with
29
CEC2017
10
CEC2020
optimization
benchmark
functions.
addition,
practicality
tested
real-world
feature
selection
problem
five
design
Results
are
compared
against
number
classical
algorithms
using
statistical
metrics,
convergence
analysis,
box
plots,
Wilcoxon
rank
sum
test.
case
composite
test
functions
F21-F30,
wins
70%
cases,
whereas
rest
functions,
generates
good
results
65%
cases.
proposed
produces
performance
55%
45%
generated
competitive
results.
On
other
hand,
problems,
among
algorithms.
problem,
also
showed
competitiveness
observations
problems
show
superiority
robustness
baseline
metaheuristics.
It
can
be
safely
concluded
improvements
suggested
proved
effective
making
enough
solve
variety
Journal of King Saud University - Computer and Information Sciences,
Journal Year:
2023,
Volume and Issue:
35(7), P. 101611 - 101611
Published: June 10, 2023
The
process
of
creating
high-quality
labeled
data
is
crucial
for
training
machine-learning
models,
but
it
can
be
a
time-consuming
and
labor-intensive
process.
Moreover,
manual
annotation
by
human
annotators
lead
to
varying
degrees
competency,
training,
experience,
which
result
in
inconsistent
labeling
arbitrary
standards.
To
address
these
challenges,
researchers
have
been
exploring
automated
methods
enhancing
testing
datasets.
This
paper
proposes
SRL-ACO,
novel
text
augmentation
framework
that
leverages
Semantic
Role
Labeling
(SRL)
Ant
Colony
Optimization
(ACO)
techniques
generate
additional
natural
language
processing
(NLP)
models.
uses
SRL
identify
the
semantic
roles
words
sentence
ACO
new
sentences
preserve
roles.
SRL-ACO
enhance
accuracy
NLP
models
generating
without
requiring
annotation.
presents
experimental
results
demonstrating
effectiveness
on
seven
classification
datasets
sentiment
analysis,
toxic
detection
sarcasm
identification.
show
improves
performance
classifier
different
tasks.
These
demonstrate
has
potential
quality
quantity
various