Online
time
series
play
a
crucial
role
in
the
analysis
and
management
of
massive
amounts
data.
These
capture
data
points
chronologically
according
to
their
acquisition
time.
Detecting
anomalies
(outliers)
these
is
for
understanding
patterns
making
informed
decisions.
This
work
exposes
various
techniques
from
literature
online
anomaly
detection,
categorises
them
into
statistical
techniques.
The
paper
shows
several
applications
methods,
machine
learning,
hybrid
methods
that
leverage
advantages
both
deep
learning
Furthermore,
ensembles
are
exposed
as
an
efficient
technique
used
with
mentioned
models
detection
series.
discusses
challenges
associated
temporal
correlation,
including
need
effective
visualisation
tools
such
DeepVats.
By
providing
overview
existing
applications,
this
aims
contribute
advancement
provide
insight
future
research
field.
Fire,
Год журнала:
2024,
Номер
7(12), С. 482 - 482
Опубликована: Дек. 18, 2024
The
increasing
frequency
and
intensity
of
wildfires
highlight
the
need
to
develop
more
efficient
tools
for
firefighting
management,
particularly
in
field
wildfire
spread
prediction.
Classical
models
have
relied
on
mathematical
empirical
approaches,
which
trouble
capturing
complexity
fire
dynamics
suffer
from
poor
flexibility
static
assumptions.
emergence
machine
learning
(ML)
and,
specifically,
deep
(DL)
has
introduced
new
techniques
that
significantly
enhance
prediction
accuracy.
ML
models,
such
as
support
vector
machines
ensemble
use
tabular
data
points
identify
patterns
predict
behavior.
However,
these
often
struggle
with
dynamic
nature
wildfires.
In
contrast,
DL
convolutional
neural
networks
(CNNs)
recurrent
(CRNs),
excel
at
handling
spatiotemporal
complexities
data.
CNNs
are
effective
analyzing
spatial
satellite
imagery,
while
CRNs
suited
both
sequential
data,
making
them
highly
performant
predicting
This
paper
presents
a
systematic
review
recent
developed
prediction,
detailing
commonly
used
datasets,
improvements
achieved,
limitations
current
methods.
It
also
outlines
future
research
directions
address
challenges,
emphasizing
potential
play
an
important
role
management
mitigation
strategies.
Deep
Learning
algorithms
are
considered
"black-box"
because
it
is
not
possible
to
analyse
how
they
find
the
final
result.
This
greatly
limits
their
application
in
several
domains,
especially
fields
like
medicine,
where
errors
can
harm
patients.
To
overcome
this
limitation,
explainable
AI
techniques
have
been
developed
that
allow
us
understand
features
of
input
relevant
system
Most
authors
do
pay
enough
attention
techniques,
creating
very
basic
and
uninformative
representations.
For
reason,
we
different
heatmap-based
eXplainable
for
medical
problems
related
chest
x-rays
classification,
depending
on
classification
problem:
binary
mutilabel.
In
our
methodology,
divide
into
two
groups
address
explainability
Artificial
Intelligence
applied
show
five
representative
examples
visualisation
techniques.
International Journal of Digital Earth,
Год журнала:
2024,
Номер
17(1)
Опубликована: Окт. 30, 2024
The
spatial
and
temporal
variations
of
net
surface
radiation
(Rn)
are
critical
for
comprehending
ecological
environments.
Nonetheless,
the
intricate
interplay
among
Rn
dynamics,
vegetation
growth,
climate,
natural
factors
remains
inadequately
elucidated.
In
this
study,
we
estimated
based
on
Landsat
data
ERA5
meteorological
in
Google
Earth
Engine
(GEE)
platform,
which
closely
matched
observable
distribution
(R2
=
0.96),
with
an
average
growth
rate
0.15
MJ
m−2
mth−1.
Trend
analyses
autocorrelation
were
used
to
explore
changes
from
2000
2020,
global
Moran's
index
was
found
exceed
0.76,
fluctuating
increases,
showing
a
highly
positive
Rn.
Local
I
predominantly
fell
into
two
categories:
'High-High'
'Low-Low',
first
increasing
range
latter
decreasing.
Combining
GeoDetector
PLS-SEM
analyses,
temperature
emerge
as
predominant
drivers
variation
within
study
area,
each
contributing
more
than
17%
change.
Furthermore,
interactions
between
any
typically
exhibits
nonlinear
enhancement.
underscores
influence
climate
Rn,
other
indirectly
affecting
by
influencing
growth.
Research Square (Research Square),
Год журнала:
2024,
Номер
unknown
Опубликована: Дек. 4, 2024
Abstract
Time
series
are
essential
for
modelling
a
lot
of
activities
such
as
software
behavior,
heart
beats
per
time,
business
processes.
The
analysis
the
data
can
prevent
errors,
boost
profits,
and
improve
understanding
behaviors.
Among
many
techniques
available,
we
find
Deep
Learning
Data
Mining
techniques.
In
Mining,
distance
matrices
between
subsequences
(similarity
matrices,
recurrence
plots)
have
already
shown
their
potential
on
fast
large-scale
time
behavior
analysis.
Learning,
there
exists
different
tools
analyzing
models
embedding
space
getting
insights
behavior.
DeepVATS
is
tool
large
that
allows
visual
interaction
within
(latent
space)
original
data.
training
model
may
result
use
computational
resources,
resulting
in
lack
interactivity.
To
solve
this
issue,
integrate
plots
tool.
incorporation
these
with
associated
downsampling
makes
more
efficient
user-friendly
first
quick
data,
achieving
runtimes
reductions
up
to
\(10^4\)
seconds,
allowing
preliminary
datasets
7M
elements.
Also,
us
detect
trends,
extending
its
capabilities.
new
functionality
tested
three
cases:
M-Toy
synthetic
dataset
anomaly
detection,
S3
trend
detection
real-world
Pulsus
Paradoxus
checking.
Fire and Materials,
Год журнала:
2024,
Номер
unknown
Опубликована: Дек. 27, 2024
ABSTRACT
The
occurrence
of
wildfire
disasters
can
easily
trigger
tripping
in
overhead
transmission‐line,
thereby
posing
a
significant
threat
to
the
safe
and
stable
operation
power
system.
In
order
enhance
prevention
control
capability
risk
assessment
method
based
on
improved
analytic
hierarchy
process
(AHP)
is
proposed.
First,
main
factors
are
explored,
indicator
system
for
transmission‐line
constructed.
We
propose
novel
runaway
coefficient
fire
assessing
impact
sources
disaster.
Secondly,
mutual
information
used
avoid
subjective
arbitrariness
AHP
improve
reliability
each
index
weight.
results
show
that
about
82.14%
new
events
2023
Fujian
(China)
located
medium‐,
high‐,
very‐high‐risk
areas,
demonstrating
effectiveness
proposed
method.
This
methodology
offers
foundation
mitigate
wildfire.
