Applied Mathematics and Nonlinear Sciences,
Год журнала:
2025,
Номер
10(1)
Опубликована: Янв. 1, 2025
Abstract
The
rapid
expansion
of
cross-national
e-commerce
has
brought
significant
opportunities
and
challenges
in
understanding
diverse
consumer
behavior.
This
study
introduces
an
innovative
framework
combining
the
XLSTM
(Extended
Long
Short-Term
Memory)
model
with
K-means
clustering
to
analyze
user
behavior
optimize
conversion
rates
on
global
platforms.
extends
traditional
LSTM
models
by
incorporating
multi-dimensional
cell
states,
attention
mechanisms,
improved
memory
capabilities,
enabling
it
effectively
capture
complex
temporal
cross-cultural
patterns.
integration
enhances
process
providing
high-quality
embeddings
that
lead
well-defined
stable
clusters.
Through
comprehensive
evaluations,
combined
approach
demonstrates
superior
performance
across
key
metrics,
including
Silhouette
Score,
Davies-Bouldin
Index
(DBI),
Adjusted
Rand
(ARI),
compared
standalone
algorithms
LSTM-based
methods.
Feature
importance
analysis
further
identifies
coupon
usage,
visit
frequency,
product
category
interest
as
most
influential
factors
purchase
decisions.
findings
highlight
potential
this
methodology
improve
engagement
marketing
strategies
for
Physics-informed
neural
networks
(PINNs)
represent
a
significant
advancement
at
the
intersection
of
machine
learning
and
physical
sciences,
offering
powerful
framework
for
solving
complex
problems
governed
by
laws.
This
survey
provides
comprehensive
review
current
state
research
on
PINNs,
highlighting
their
unique
methodologies,
applications,
challenges,
future
directions.
We
begin
introducing
fundamental
concepts
underlying
motivation
integrating
physics-based
constraints.
then
explore
various
PINN
architectures
techniques
incorporating
laws
into
network
training,
including
approaches
to
partial
differential
equations
(PDEs)
ordinary
(ODEs).
Additionally,
we
discuss
primary
challenges
faced
in
developing
applying
such
as
computational
complexity,
data
scarcity,
integration
Finally,
identify
promising
Overall,
this
seeks
provide
foundational
understanding
PINNs
within
rapidly
evolving
field.
Journal of Chemical Information and Modeling,
Год журнала:
2025,
Номер
unknown
Опубликована: Янв. 21, 2025
Skin
corrosion
assessment
is
an
essential
toxicity
end
point
that
addresses
safety
concerns
for
topical
dosage
forms
and
cosmetic
products.
Previously,
skin
assessments
required
animal
testing;
however,
differences
in
architecture
ethical
regarding
models
have
fostered
the
advancement
of
alternative
methods
such
as
silico
vitro
models.
This
study
aimed
to
develop
deep
learning
(DL)
based
on
recurrent
neural
networks
(RNNs)
classifying
chemical
compounds
language
notation,
molecular
substructure,
physicochemical
properties,
a
combination
these
three
properties
called
conjoint
fingerprints.
Simple
RNN,
long
short-term
memory,
bidirectional
memory
(BiLSTM),
gated
units,
units
models,
along
with
11
features,
were
employed
generate
55
RNN-based
Applicability
domain
permutation
importance
analysis
exploited
additional
trustable
prediction
explanation
ability
respectively.
Our
findings
indicate
BiLSTM
features
MACCS
keys
descriptors
most
effective
model
84.3%
accuracy,
89.8%
area
under
curve,
57.6%
Matthews
correlation
coefficient
external
test
performance.
Furthermore,
our
accurately
predicted
all
new
unseen
beyond
set,
highlighting
prominent
classification
performance
compared
existing
finding
will
contribute
utilization
DL
characteristics
structure
enhance
model's
predictive
capability
assessment.
IEEE Access,
Год журнала:
2024,
Номер
12, С. 20540 - 20558
Опубликована: Янв. 1, 2024
Diabetic
retinopathy
(DR)
is
a
microvascular
disease
that
associated
with
diabetes
mellitus.
DR
can
cause
irreversible
vision
loss
and
blindness.
classification,
is,
early
diagnosis
accurate
grading,
critical
for
protection
immediate
treatment.
Deep
learning-based
automated
systems
led
to
significant
expectations
classification
based
on
fundus
images
several
advantages.
In
the
past
years,
many
outstanding
studies
in
this
area
have
been
conducted
review
articles
published.
However,
new
trends
future
directions
are
need
further
analyzed.
Thus,
we
carefully
included
read
94
related
published
from
2018
2023
through
Web
of
Science,
PubMed,
Scopus,
IEEE
Xplore.
From
review,
found
transfer
learning
has
used
as
an
strategy
overcoming
issue
limited
data
resources
support
analysis.
CNN
models
ResNet
VGGNet
layers
tens
or
even
hundreds
most
popular
frameworks
classification.
The
APTOS
2019
EyePACS
widely
datasets
addition,
some
lightweight
DL
architectures
like
SqueezeNet
MobileNet
proposed
tasks,
especially
computational
capabilities.
Although
deep
achieved
surpassed
human-level
accuracy
there
still
long
way
go
real
clinical
workflows.
Further
improvements
model
interpretability,
trustworthiness
ophthalmologists,
cost-effective
reliable
screening
needed.
Artificial Intelligence Review,
Год журнала:
2024,
Номер
57(11)
Опубликована: Сен. 18, 2024
Abstract
In
recent
years,
Advanced
Persistent
Threat
(APT)
attacks
on
network
systems
have
increased
through
sophisticated
fraud
tactics.
Traditional
Intrusion
Detection
Systems
(IDSs)
suffer
from
low
detection
accuracy,
high
false-positive
rates,
and
difficulty
identifying
unknown
such
as
remote-to-local
(R2L)
user-to-root
(U2R)
attacks.
This
paper
addresses
these
challenges
by
providing
a
foundational
discussion
of
APTs
the
limitations
existing
methods.
It
then
pivots
to
explore
novel
integration
deep
learning
techniques
Explainable
Artificial
Intelligence
(XAI)
improve
APT
detection.
aims
fill
gaps
in
current
research
thorough
analysis
how
XAI
methods,
Shapley
Additive
Explanations
(SHAP)
Local
Interpretable
Model-agnostic
(LIME),
can
make
black-box
models
more
transparent
interpretable.
The
objective
is
demonstrate
necessity
explainability
propose
solutions
that
enhance
trustworthiness
effectiveness
models.
offers
critical
approaches,
highlights
their
strengths
limitations,
identifies
open
issues
require
further
research.
also
suggests
future
directions
combat
evolving
threats,
paving
way
for
effective
reliable
cybersecurity
solutions.
Overall,
this
emphasizes
importance
enhancing
performance
systems.
