IEEE Access,
Journal Year:
2024,
Volume and Issue:
12, P. 136222 - 136229
Published: Jan. 1, 2024
Aiming
to
address
the
issue
of
low
accuracy
in
industrial
network
traffic
anomaly
detection,
we
propose
an
improved
DeepFM
model
for
multi-type
detection.
The
dataset
undergoes
preprocessing,
including
encoding
and
non-string
numerical
operations.
SMOTE-ENN
algorithm
is
utilized
balance
data
through
oversampling
undersampling.
employed
extract
linear,
non-linear,
temporal
features
from
data.
These
are
then
fed
into
detector
classifier
constructed
based
on
Softmax
achieve
high-performance
detection
attacks.
effectiveness
verified
using
UNSW-NB15
dataset,
with
experimental
results
demonstrating
a
0.95
DoS
attacks,
0.94
Fuzzers
0.92
Worms
significantly
surpassing
other
algorithms,
which
confirms
effective
utilization
proposed
IEEE Access,
Journal Year:
2024,
Volume and Issue:
12, P. 61642 - 61666
Published: Jan. 1, 2024
The
Internet
of
Things
(IoT)
represents
a
dynamic
infrastructure,
leveraging
sensing
and
network
communication
technology
to
establish
ubiquitous
connectivity
among
people,
machines,
objects.
Due
its
end
devices'
limited
computing
resources
storage
space,
it
is
not
feasible
merely
transpose
traditional
internet
security
technologies
directly
IoT
endpoints.
Maintaining
while
concurrently
ensuring
performance
particularly
challenging
endeavor.
This
paper
provides
review
key
agreements
authentication
protocols
pivotal
the
IoT.
First,
this
survey
discusses
applications
that
need
agreement
strengthen
their
current
research
on
these
application
fields.
Subsequently,
engages
in
an
in-depth
exploration
phase
involved
scheme
agreement,
including
examination
cryptographic
techniques
employed
within
processes.
also
thoroughly
studies
scheme's
services,
potential
attacks,
formal
analysis
informal
ensure
resilience
against
such
threats.
study
aims
provide
profound
understanding
recent
applications.
It
strives
contribute
towards
strengthening
systems
for
applications,
sustainability
face
evolving
Future Internet,
Journal Year:
2025,
Volume and Issue:
17(1), P. 25 - 25
Published: Jan. 8, 2025
The
rapid
evolution
of
technologies
such
as
the
Internet
Things
(IoT),
5G,
and
cloud
computing
has
exponentially
increased
complexity
cyber
attacks.
Modern
Intrusion
Detection
Systems
(IDSs)
must
be
capable
identifying
not
only
frequent,
well-known
attacks
but
also
low-frequency,
subtle
intrusions
that
are
often
missed
by
traditional
systems.
challenge
is
further
compounded
fact
most
IDS
rely
on
black-box
machine
learning
(ML)
deep
(DL)
models,
making
it
difficult
for
security
teams
to
interpret
their
decisions.
This
lack
transparency
particularly
problematic
in
environments
where
quick
informed
responses
crucial.
To
address
these
challenges,
we
introduce
XI2S-IDS
framework—an
Explainable,
Intelligent
2-Stage
System.
framework
uniquely
combines
a
two-stage
approach
with
SHAP-based
explanations,
offering
improved
detection
interpretability
low-frequency
Binary
classification
conducted
first
stage
followed
multi-class
second
stage.
By
leveraging
SHAP
values,
enhances
decision-making,
allowing
analysts
gain
clear
insights
into
feature
importance
model’s
rationale.
Experiments
UNSW-NB15
CICIDS2017
datasets
demonstrate
significant
improvements
performance,
notable
reduction
false
negative
rates
attacks,
while
maintaining
high
precision,
recall,
F1-scores.
e-Prime - Advances in Electrical Engineering Electronics and Energy,
Journal Year:
2024,
Volume and Issue:
9, P. 100673 - 100673
Published: July 5, 2024
The
issue
of
network
security
is
an
important
and
delicate
when
it
comes
to
the
privacy
organizations
individuals,
especially
sensitive
information
transmitted
across
these
networks.
importance
intrusion
detection
systems,
which
a
very
component
protecting
reducing
damage
resulting
from
attacks
penetrations
has
increased
due
adoption
most
recent
regulations
on
advanced
web
services,
whether
government
banking
e-mail,
or
e-marketing.
goal
this
paper
construct
system
using
deep
learning
algorithms
based
new
dataset
named
CICIoT2023.
proposed
model
addresses
challenges
associated
with
datasets
in
terms
high
dimensionality
by
adopting
methods
reduce
their
size
improve
efficiency.
A
clustering
technique
for
method
combination
between
optimization
algorithm
static
tools
was
proposed.
evaluated
determine
its
efficiency
several
evaluation
measures.
results
show
that
comparison
earlier
research
conducted
same
datasets,
suggested
performs
better
attack
detection.
As
result,
offers
level
trust.
PeerJ Computer Science,
Journal Year:
2025,
Volume and Issue:
11, P. e2682 - e2682
Published: Feb. 25, 2025
As
the
world
grapples
with
pandemics
and
increasing
stress
levels
among
individuals,
heart
failure
(HF)
has
emerged
as
a
prominent
cause
of
mortality
on
global
scale.
The
most
effective
approach
to
improving
chances
individuals'
survival
is
diagnose
this
condition
at
an
early
stage.
Researchers
widely
utilize
supervised
feature
selection
techniques
alongside
conventional
standalone
machine
learning
(ML)
algorithms
achieve
goal.
However,
these
approaches
may
not
consistently
demonstrate
robust
performance
when
applied
data
that
they
have
encountered
before,
struggle
discern
intricate
patterns
within
data.
Hence,
we
present
Multi-objective
Stacked
Enable
Hybrid
Model
(MO-SEHM),
aims
find
out
best
subsets
numerous
different
sets,
considering
multiple
objectives.
(SEHM)
plays
role
classifier
integrates
multi-objective
method,
Non-dominated
Sorting
Genetic
Algorithm
II
(NSGA-II).
We
employed
HF
dataset
from
Faisalabad
Institute
Cardiology
(FIOC)
evaluated
six
ML
models,
including
SEHM
without
NSGA-II
for
experimental
purposes.
Pareto
front
(PF)
demonstrates
our
introduced
MO-SEHM
surpasses
other
obtaining
94.87%
accuracy
nine
relevant
features.
Finally,
Local
Interpretable
Model-agnostic
Explanations
(LIME)
explain
reasons
individual
outcomes,
which
makes
model
transparent
patients
stakeholders.
Wiley Interdisciplinary Reviews Data Mining and Knowledge Discovery,
Journal Year:
2025,
Volume and Issue:
15(2)
Published: March 28, 2025
ABSTRACT
As
the
Internet
of
Things
(IoT)
continues
expanding
its
footprint
across
various
sectors,
robust
security
systems
to
mitigate
associated
risks
are
more
critical
than
ever.
Intrusion
Detection
Systems
(IDS)
fundamental
in
safeguarding
IoT
infrastructures
against
malicious
activities.
This
systematic
review
aims
guide
future
research
by
addressing
six
pivotal
questions
that
underscore
development
advanced
IDS
tailored
for
environments.
Specifically,
concentrates
on
applying
machine
learning
(ML)
and
deep
(DL)
technologies
enhance
capabilities.
It
explores
feature
selection
methodologies
aimed
at
developing
lightweight
solutions
both
effective
efficient
scenarios.
Additionally,
assesses
different
datasets
balancing
techniques,
which
crucial
training
models
perform
accurately
reliably.
Through
a
comprehensive
analysis
existing
literature,
this
highlights
significant
trends,
identifies
current
gaps,
suggests
studies
optimize
frameworks
ever‐evolving
landscape.
IEEE Transactions on Consumer Electronics,
Journal Year:
2024,
Volume and Issue:
70(1), P. 3951 - 3959
Published: Feb. 1, 2024
Deep
learning-driven
side-channel
analysis
(SCA)
is
a
promising
approach
to
analytic
profiling.
Recent
studies
have
shown
that
neural
networks
can
successfully
attack
defended
targets,
even
with
small
number
of
traces.
However,
developing
requires
fine-tuning
hyperparameters,
which
challenging
and
time-consuming,
especially
for
complex
networks.
This
study
proposes
an
AutoSCA
framework
uses
Bayesian
optimization
automate
deep
learning
hyperparameter
tuning
SCA.
The
implemented
using
two
popular
network
architectures:
the
multi-layer
perceptron
(MLP)
convolutional
(CNN).
improves
performance
measurements,
has
potential
applications
in
6G
communication-based
mobile
devices.
was
trained
evaluated
ASCAD
CHES
CTF
datasets.
experimental
results
showed
CNN-based
outperformed
MLP-based
other
state-of-the-art
models,
terms
low
time
complexity
higher
accuracy.
Results
suggest
effective
regardless
dataset,
architecture,
or
type
leaky
prototype
defeating
contemporary
attacks.
Applying
against
attacks
consumer
electronics
significantly
enhance
security
user
data
privacy
increasingly
connected.