Intelligent Systems with Applications,
Год журнала:
2023,
Номер
18, С. 200202 - 200202
Опубликована: Фев. 15, 2023
In
this
paper,
stock
price
data
has
been
predicted
using
several
state-of-the-art
methodologies
such
as
stochastic
models,
machine
learning
techniqus,
and
deep
algorithms.
An
efficient
decomposition
method
resonating
with
these
Machine
Intelligence
(MI)
models
embedded
boosting
ensemble
method.
Finally
a
Model
Confidence
Set
(MCS)
based
algorithm
proposed
for
forecasting
data.
Complete
Ensemble
Empirical
Mode
Decomposition
Adaptive
Noise
(CEEMDAN)
decomposed
orthogonal
subseries
have
Random
Forests
(RFs).
Then
Kernel
Ridge
Regression
(KRR)
model
is
used
to
combine
those
predictions
form
hybrid
predictor.
addition,
improvement
in
prediction
performance
observed
kernel
functions.
Boosting
(AdaBoost)
found
stimulating
accuracy
of
Long
Short-Term
Memory
(LSTM)
Gated
Recurrent
Unit
(GRU)
models.
CEEMDAN
also
increased
the
AdaBoost.
Nevertheless,
combination
forecasts
from
various
good
approach
improving
result.
Despite
optimizing
weights
all
heuristic
MCS-based
snuffing
least
important
prior
averaging
conceded
potent
approach.
MCS
rescinds
insignificant
on
out-of-sample
or
in-sample
equally
average
superior
The
compared
existing
standalone
techniques
validation
measures.
However,
Support
Vector
(CCEMDAN_SVR)
be
best
predictor
current
scenario.
Journal of Computational Science,
Год журнала:
2023,
Номер
69, С. 102010 - 102010
Опубликована: Март 31, 2023
The
Dynamic
Hunting
Leadership
(DHL)
algorithm
is
an
innovative
heuristic
technique
that
draws
inspiration
from
nature
to
find
almost
optimal
solutions
for
various
optimization
problems.
It
consists
of
four
variants,
each
highlighting
distinct
leadership
strategies
guide
the
hunting
process.
development
was
based
on
realization
effective
during
process
can
significantly
improve
its
efficacy.
concept
behind
these
methods
dynamically
modify
number
leaders,
which
enhance
algorithm's
performance.
stability
DHL
variants
in
exploring
unknown
area
search
space
and
exploitation
phases
compared,
advantages
exploration
or
ability
different
are
discussed.
Moreover,
results
compared
with
more
than
twenty
well-known
algorithms.
efficacy
proposed
algorithms
discovering
near-optimal
tested
across
several
real-world
applications,
outcomes
demonstrate
outperforms
other
competing
distinctiveness
identify
global
minimum
benchmark
problems
superior
performance
enhancing
objective
value
welded
beam
design
problem
tension–compression
spring
problem,
surpassing
values
achieved
by
algorithms'
also
allocation
distributed
generation
(DG)
energy
storage
system
(ESS)
balanced
electrical
distribution
systems.
show
all
obtained
problem.
For
control
strategy
voltage
regulators
three-phase
unbalanced
power
systems,
improved
35.7%
best
found
Based
analysis
comparison
convergence
behavior
benchmarks
problems,
method
proves
be
reliable
method.
International journal of intelligent engineering and systems,
Год журнала:
2023,
Номер
16(3), С. 345 - 361
Опубликована: Май 1, 2023
This
research
presents
a
novel
hybrid
sampling
technique,
implemented
at
the
data
level,
to
effectively
address
imbalanced
and
noisy
in
classification
processes.The
proposed
technique
expertly
combines
two
established
methods,
namely,
random
over
(ROS)
neighbourhood
cleaning
rule
(NCL)
approaches,
tackle
imbalance
noise
issues,
respectively.The
study
carried
out
an
empirical
evaluation
of
approach
using
crowdsourced
text
that
primarily
emphasized
triple
bottom
line
(TBL)
dimension
smart
social,
economic,
environmental
city.The
used
long
short-term
memory
(LSTM),
convolutional
neural
networks
(CNN),
CNN-LSTM
models
validate
efficacy
compare
its
performance
with
other
existing
including
ROS
oversampling,
NCL
undersampling,
synthetic
minority
&
tomek
links
(SMOTE-Tomek),
oversampling
edited
nearest
neighbours
(SMOTE-ENN)
sampling.The
results
are
impressive,
ROS-NCL
achieving
high
accuracy
rates
across
all
three
models,
97.71%,
98.01%,
98.11%,
respectively.This
provides
robust
effective
solution
for
handling
impure
holds
great
promise
identifying
complex
patterns
real-world
problems.
Intelligent Systems with Applications,
Год журнала:
2023,
Номер
18, С. 200202 - 200202
Опубликована: Фев. 15, 2023
In
this
paper,
stock
price
data
has
been
predicted
using
several
state-of-the-art
methodologies
such
as
stochastic
models,
machine
learning
techniqus,
and
deep
algorithms.
An
efficient
decomposition
method
resonating
with
these
Machine
Intelligence
(MI)
models
embedded
boosting
ensemble
method.
Finally
a
Model
Confidence
Set
(MCS)
based
algorithm
proposed
for
forecasting
data.
Complete
Ensemble
Empirical
Mode
Decomposition
Adaptive
Noise
(CEEMDAN)
decomposed
orthogonal
subseries
have
Random
Forests
(RFs).
Then
Kernel
Ridge
Regression
(KRR)
model
is
used
to
combine
those
predictions
form
hybrid
predictor.
addition,
improvement
in
prediction
performance
observed
kernel
functions.
Boosting
(AdaBoost)
found
stimulating
accuracy
of
Long
Short-Term
Memory
(LSTM)
Gated
Recurrent
Unit
(GRU)
models.
CEEMDAN
also
increased
the
AdaBoost.
Nevertheless,
combination
forecasts
from
various
good
approach
improving
result.
Despite
optimizing
weights
all
heuristic
MCS-based
snuffing
least
important
prior
averaging
conceded
potent
approach.
MCS
rescinds
insignificant
on
out-of-sample
or
in-sample
equally
average
superior
The
compared
existing
standalone
techniques
validation
measures.
However,
Support
Vector
(CCEMDAN_SVR)
be
best
predictor
current
scenario.
