A high-precision crown control strategy for hot-rolled electric steel using theoretical model-guided BO-CNN-BiLSTM framework
Chunning Song,
No information about this author
Jianguo Cao,
No information about this author
Qiufang Zhao
No information about this author
et al.
Applied Soft Computing,
Journal Year:
2024,
Volume and Issue:
152, P. 111203 - 111203
Published: Jan. 5, 2024
Language: Английский
Comparative analysis of advanced deep learning models for predicting evapotranspiration based on meteorological data in bangladesh
Sourov Paul,
No information about this author
Syeda Zehan Farzana,
No information about this author
Saikat Das
No information about this author
et al.
Environmental Science and Pollution Research,
Journal Year:
2024,
Volume and Issue:
unknown
Published: Oct. 4, 2024
Language: Английский
Physics-informed machine learning for HSV performance degradation prediction in water hydraulic manipulator
Ruidong Hong,
No information about this author
Songlin Nie,
No information about this author
Hui Ji
No information about this author
et al.
Reliability Engineering & System Safety,
Journal Year:
2025,
Volume and Issue:
unknown, P. 111106 - 111106
Published: April 1, 2025
Language: Английский
Bearing Fault Diagnosis Method Based on Improved VMD and Parallel Hybrid Neural Network
Wuyi Chen,
No information about this author
Huafeng Cai,
No information about this author
Sun Qiu
No information about this author
et al.
Applied Sciences,
Journal Year:
2025,
Volume and Issue:
15(8), P. 4430 - 4430
Published: April 17, 2025
In
order
to
combat
the
difficulty
of
fault
feature
extraction
and
recognition
in
field
bearing
diagnosis,
a
diagnosis
method
based
on
improved
variational
mode
decomposition
(VMD)
parallel
hybrid
neural
network
is
proposed,
which
combines
reweighted
kurtosis
(RK)
with
variable
uses
as
evaluation
index
select
times
decomposition,
while
removing
part
interference
signal
retaining
its
impact
characteristics.
Afterwards,
processed
data
set
brought
into
model
global
average
pooling
layer
(GAP)
for
extraction,
fusion,
classification.
The
can
extract
features
more
comprehensively
improve
accuracy
speed
up
training
testing.
Experiments
Xian
Jiao
tong
University
(XJTU)
Case
Western
Reserve
(CWRU)
public
sets
show
that
reaches
99.72%
99.73%,
respectively,
indicating
has
good
better
performance
compared
other
models.
Language: Английский
Advancements in Healthcare Medical Imaging through SHO optimized CNN
Procedia Computer Science,
Journal Year:
2025,
Volume and Issue:
258, P. 4128 - 4135
Published: Jan. 1, 2025
Language: Английский
Application of a grey wolf optimization-enhanced convolutional neural network and bidirectional gated recurrent unit model for credit scoring prediction
PLoS ONE,
Journal Year:
2025,
Volume and Issue:
20(5), P. e0322225 - e0322225
Published: May 27, 2025
With
the
digital
transformation
of
financial
industry,
credit
score
prediction,
as
a
key
component
risk
management,
faces
increasingly
complex
challenges.
Traditional
scoring
methods
often
have
difficulty
in
fully
capturing
characteristics
large-scale,
high-dimensional
data,
resulting
limited
prediction
performance.
To
address
these
issues,
this
paper
proposes
model
that
combines
CNNs
and
BiGRUs,
uses
GWO
algorithm
for
hyperparameter
tuning.
CNN
performs
well
feature
extraction
can
effectively
capture
patterns
customer
historical
behaviors,
while
BiGRU
is
good
at
handling
time
dependencies,
which
further
improves
accuracy
model.
The
introduced
to
improve
overall
performance
by
optimizing
parameters.
Experimental
results
show
CNN-BiGRU-GWO
proposed
on
multiple
public
datasets,
significantly
improving
efficiency
prediction.
On
LendingClub
loan
dataset,
MAE
15.63,
MAPE
4.65%,
RMSE
3.34,
MSE
12.01,
are
64.5%,
68.0%,
21.4%,
52.5%
lower
than
traditional
method
plawiak
44.07,
14.51%,
4.25,
25.29,
respectively.
In
addition,
compared
with
methods,
also
shows
stronger
advantages
adaptability
generalization
ability.
By
integrating
advanced
technologies,
not
only
provides
an
innovative
technical
solution
but
valuable
insights
into
application
deep
learning
field,
making
up
shortcomings
existing
demonstrating
its
potential
wide
management.
Language: Английский
Evaluating smart grid investment drivers and creating effective policies via a fuzzy multi-criteria approach
Renewable and Sustainable Energy Reviews,
Journal Year:
2024,
Volume and Issue:
208, P. 115052 - 115052
Published: Oct. 31, 2024
Language: Английский
Deep learning approaches for short-crop reference evapotranspiration estimation: a case study in Southeastern Australia
Uaktho Baishnab,
No information about this author
Md. Sahadat Hossen Sajib,
No information about this author
Ashraful Islam
No information about this author
et al.
Earth Science Informatics,
Journal Year:
2024,
Volume and Issue:
18(1)
Published: Dec. 4, 2024
Language: Английский
Enterprise financial sharing and risk identification model combining recurrent neural networks with transformer model supported by blockchain
Yang Wu
No information about this author
Heliyon,
Journal Year:
2024,
Volume and Issue:
10(12), P. e32639 - e32639
Published: June 1, 2024
The
objective
of
this
study
is
to
investigate
methodologies
concerning
enterprise
financial
sharing
and
risk
identification
mitigate
concerns
associated
with
the
safeguarding
data.
Initially,
analysis
examines
security
vulnerabilities
inherent
in
conventional
information
practices.
Subsequently,
blockchain
technology
introduced
transition
various
entity
nodes
within
centralized
networks
into
a
decentralized
framework,
culminating
formulation
blockchain-based
model
for
data
sharing.
Concurrently,
integrates
Bi-directional
Long
Short-Term
Memory
(BiLSTM)
algorithm
transformer
model,
presenting
an
referred
as
BiLSTM-fused
model.
This
amalgamates
multimodal
sequence
modeling
comprehensive
understanding
both
textual
visual
It
stratifies
values
levels
1
5,
where
level
signifies
most
favorable
condition,
followed
by
relatively
good
(level
2),
average
3),
high
4),
severe
5).
Subsequent
construction,
experimental
conducted,
revealing
that,
comparison
Byzantine
Fault
Tolerance
(BFT)
mechanism,
proposed
achieves
throughput
exceeding
80
node
count
146.
Both
message
leakage
packet
loss
rates
remain
below
10
%.
Moreover,
when
juxtaposed
recurrent
neural
(RNNs)
algorithm,
demonstrates
accuracy
surpassing
94
%,
AUC
value
0.95,
reduction
time
required
approximately
s.
Consequently,
facilitates
more
precise
efficient
potential
risks,
thereby
furnishing
crucial
support
management
strategic
decision-making
endeavors.
Language: Английский
Intelligent Asset Allocation Portfolio Division and Recommendation
Liang Cai,
No information about this author
Zhixin Wu
No information about this author
Journal of Organizational and End User Computing,
Journal Year:
2024,
Volume and Issue:
36(1), P. 1 - 23
Published: Sept. 16, 2024
With
the
continuous
development
of
financial
markets,
intelligent
asset
allocation
has
become
a
topic
great
concern
in
investment
field.
However,
traditional
methods
often
face
difficulties
grasping
relationship
between
diversity,
risk
and
return,
which
limits
its
application
complex
market
environments.
To
solve
this
problem,
study
introduces
deep
learning
knowledge
graphs
proposes
an
model.
Our
model
makes
full
use
advantages
Knowledge
Graph
Embedding
Model
(KGE),
LSTM,
Genetic
Algorithm
(GA)
to
build
multi-level
multi-dimensional
KGE
helps
capture
relationships
different
assets,
LSTM
is
used
learn
key
patterns
historical
portfolio
performance,
GA
finds
optimal
combination
by
simulating
natural
selection
genetic
mechanisms.
Experimental
findings
indicate
that
our
demonstrated
substantial
improvements
across
various
performance
metrics
outperforms
conventional
approaches.
Language: Английский