Fractal and Fractional,
Journal Year:
2024,
Volume and Issue:
8(12), P. 720 - 720
Published: Dec. 5, 2024
Aiming
at
the
uncertainty
problem
caused
by
time-varying
modeling
parameters
associated
with
ship
speed
in
course
tracking
control
of
underactuated
surface
vessels
(USVs),
this
paper
proposes
a
algorithm
based
on
dynamic
neural
fuzzy
model
(DNFM).
The
DNFM
simultaneously
adjusts
structure
and
during
learning
fully
approximates
inverse
dynamics
ships.
Online
identification
lays
foundation
for
motion
control.
trained
DNFM,
serving
as
an
controller,
is
connected
parallel
fractional-order
PIλDμ
controller
to
be
used
ship’s
course.
Moreover,
weights
can
further
adjusted
tracking.
Taking
actual
data
5446
TEU
large
container
ship,
simulation
experiments
are
conducted,
respectively,
tracking,
under
wind
wave
interferences,
comparison
five
different
controllers.
This
proposed
overcome
influence
parameters,
desired
quickly
effectively.
Electronics,
Journal Year:
2025,
Volume and Issue:
14(9), P. 1879 - 1879
Published: May 5, 2025
The
Multiple
Random
Empirical
Kernel
Learning
Machine
(MREKLM)
typically
generates
multiple
empirical
feature
spaces
by
selecting
a
limited
group
of
samples,
which
helps
reduce
training
duration.
However,
MREKLM
does
not
incorporate
data
distribution
information
during
the
projection
process,
leading
to
inconsistent
performance
and
issues
with
reproducibility.
To
address
this
limitation,
we
introduce
within-class
scatter
matrix
that
leverages
resulting
in
development
Fast
Incorporating
Data
Distribution
Information
(FMEKL-DDI).
This
approach
enables
algorithm
sample
projection,
improving
decision
boundary
enhancing
classification
accuracy.
further
minimize
selection
time,
employ
border
point
technique
utilizing
locality-sensitive
hashing
(BPLSH),
efficiently
picking
samples
for
space
development.
experimental
results
from
various
datasets
demonstrate
FMEKL-DDI
significantly
improves
accuracy
while
reducing
duration,
thereby
providing
more
efficient
strong
generalization
performance.
Electronics,
Journal Year:
2024,
Volume and Issue:
13(10), P. 1930 - 1930
Published: May 15, 2024
The
high
dimensionality
of
hyperspectral
images
(HSIs)
brings
significant
redundancy
to
data
processing.
Band
selection
(BS)
is
one
the
most
commonly
used
reduction
(DR)
techniques,
which
eliminates
redundant
information
between
bands
while
retaining
a
subset
with
content
and
low
noise.
wild
horse
optimizer
(WHO)
novel
metaheuristic
algorithm
widely
for
its
efficient
search
performance,
yet
it
tends
become
trapped
in
local
optima
during
later
iterations.
To
address
these
issues,
an
enhanced
(IBSWHO)
proposed
HSI
band
this
paper.
IBSWHO
utilizes
Sobol
sequences
initialize
population,
thereby
increasing
population
diversity.
It
incorporates
Cauchy
mutation
perturb
certain
probability,
enhancing
global
capability
avoiding
optima.
Additionally,
dynamic
random
techniques
are
introduced
improve
efficiency
expand
space.
convergence
verified
on
nonlinear
test
functions
compared
state-of-the-art
optimization
algorithms.
Finally,
experiments
three
classic
datasets
conducted
classification.
experimental
results
demonstrate
that
selected
by
achieves
best
classification
accuracy
conventional
methods,
confirming
superiority
BS
method.
Remote Sensing,
Journal Year:
2024,
Volume and Issue:
16(17), P. 3155 - 3155
Published: Aug. 27, 2024
As
we
take
stock
of
the
contemporary
issue,
remote
sensing
images
are
gradually
advancing
towards
hyperspectral–high
spatial
resolution
(H2)
double-high
images.
However,
high
produces
serious
heterogeneity
and
spectral
variability
while
improving
image
resolution,
which
increases
difficulty
feature
recognition.
So
as
to
make
best
features
under
an
insufficient
number
marking
samples,
would
like
achieve
effective
recognition
accurate
classification
in
H2
In
this
paper,
a
cross-hop
graph
network
for
classification(H2-CHGN)
is
proposed.
It
two-branch
deep
extraction
geared
images,
consisting
attention
(CGAT)
multiscale
convolutional
neural
(MCNN):
CGAT
branch
utilizes
superpixel
information
filter
samples
with
relevance
designate
them
be
classified,
then
mechanism
broaden
range
convolution
obtain
more
representative
global
features.
another
branch,
MCNN
uses
dual
kernels
extract
fuse
at
various
scales
attaining
pixel-level
multi-scale
local
by
parallel
cross
connecting.
Finally,
dual-channel
utilized
fusion
elements
prominent.
This
experiment
on
classical
dataset
(Pavia
University)
datasets
(WHU-Hi-LongKou
WHU-Hi-HongHu)
shows
that
H2-CHGN
can
efficiently
competently
used
classification.
detail,
experimental
results
showcase
superior
performance,
outpacing
state-of-the-art
methods
0.75–2.16%
overall
accuracy.
