Application of novel hybrid deep learning architectures combining convolutional neural networks (CNN) and recurrent neural networks (RNN): construction duration estimates prediction considering preconstruction uncertainties
Engineering Research Express,
Год журнала:
2024,
Номер
6(3), С. 032102 - 032102
Опубликована: Авг. 20, 2024
Abstract
Construction
duration
estimation
plays
a
pivotal
role
in
project
planning
and
management,
yet
it
is
often
fraught
with
uncertainties
that
can
lead
to
cost
overruns
delays.
To
address
these
challenges,
this
review
article
proposes
three
advanced
conceptual
models
leveraging
hybrid
deep
learning
architectures
combine
Convolutional
Neural
Networks
(CNNs)
Recurrent
(RNNs)
while
considering
construction
delivery
uncertainties.
The
first
model
introduces
Spatio-Temporal
Attention
CNN-RNN
Hybrid
Model
Probabilistic
Uncertainty
Modeling,
which
integrates
attention
mechanisms
probabilistic
uncertainty
modeling
provide
accurate
estimates
of
duration,
offering
insights
into
critical
areas
uncertainty.
second
presents
Multi-Modal
Graph
Bayesian
Integration,
harnesses
multi-modal
data
sources
graph
representations
offer
comprehensive
incorporating
measures,
facilitating
informed
decision-making
optimized
resource
allocation.
Lastly,
the
third
Hierarchical
Transformer
Fuzzy
Logic
Handling,
addresses
inherent
vagueness
imprecision
by
hierarchical
spatio-temporal
transformer
architecture
fuzzy
logic
handling,
leading
more
nuanced
adaptable
management
practices.
These
represent
significant
advancements
addressing
providing
valuable
recommendations
for
future
research
industry
applications.
Moreover,
critically
examines
application
architectures,
specifically
combination
CNNs
RNNs,
predicting
at
preconstruction
stage
systems.
Язык: Английский
AD-YOLOv5: An object detection approach for key parts of sika deer based on deep learning
Computers and Electronics in Agriculture,
Год журнала:
2024,
Номер
217, С. 108610 - 108610
Опубликована: Янв. 8, 2024
Язык: Английский
Development Status, Frontier Hotspots, and Technical Evaluations in the Field of AI Music Composition Since the 21st Century: A Systematic Review
IEEE Access,
Год журнала:
2024,
Номер
12, С. 89452 - 89466
Опубликована: Янв. 1, 2024
Язык: Английский
Fault Reconstruction Method of Neural Network Observer Group for High-Speed Vehicle
Lecture notes in electrical engineering,
Год журнала:
2025,
Номер
unknown, С. 292 - 301
Опубликована: Янв. 1, 2025
Язык: Английский
Dense dynamic convolutional network for Bel canto vocal technique assessment
Scientific Reports,
Год журнала:
2025,
Номер
15(1)
Опубликована: Май 5, 2025
The
Bel
Canto
performance
is
a
complex
and
multidimensional
art
form
encompassing
pitch,
timbre,
technique,
affective
expression.
To
accurately
reflect
performer's
singing
proficiency,
it
essential
to
quantify
evaluate
their
vocal
technical
execution
precisely.
Convolutional
Neural
Networks
(CNNs),
renowned
for
robust
ability
capture
spatial
hierarchical
information,
have
been
widely
adopted
in
various
tasks,
including
audio
pattern
recognition.
However,
existing
CNNs
exhibit
limitations
extracting
intricate
spectral
features,
particularly
performance.
address
the
challenges
posed
by
features
meet
demands
objective
technique
assessment,
we
introduce
Omni-Dimensional
Dynamic
Convolution
(ODConv).
Additionally,
employ
densely
connected
layers
optimize
framework,
enabling
efficient
utilization
of
multi-scale
across
multiple
dynamic
convolution
layers.
validate
effectiveness
our
method,
conducted
experiments
on
tasks
music
classification,
acoustic
scene
sound
event
detection.
experimental
results
demonstrate
that
Dense
Network
(DDNet)
outperforms
traditional
CNN
Transformer
models,
achieving
90.11%,
73.95%,
89.31%
(Top-1
Accuracy),
41.89%
(mAP),
respectively.
Our
research
not
only
significantly
improves
accuracy
efficiency
assessment
but
also
facilitates
applications
teaching
remote
education.
Язык: Английский
Lightweight Deep-Learning Based Music Genre Classification: A Study
A. Rama,
N. Mythili,
M. P. Rajakumar
и другие.
2021 International Conference on System, Computation, Automation and Networking (ICSCAN),
Год журнала:
2023,
Номер
unknown, С. 1 - 5
Опубликована: Ноя. 17, 2023
Deep-learning
(DL)
applications
that
are
used
real-time
across
various
industries
have
gained
a
lot
of
traction
and
become
increasingly
popular,
especially
when
it
comes
to
data-driven
recommendation
systems.
This
work
aims
develop
DL
scheme
support
the
music-recommendation
system
(MS)
based
on
music
data.
The
phases
this
includes;
(i)
data
collection
signal-image
conversion
get
necessary
RGB
scale
images
from
data,
(ii)
pre-trained
feature
extraction,
(iii)
deep-features
detection
recommend
appropriate
music.
research
considered
classic-
(CL)
pop-music
(PO)
for
examination
achieved
results
evaluated
substantiate
performance
arrangement.
In
work,
procedure
is
implemented
convert
1D
signal
2D
image
then
examined
using
proposed
technique.
experimental
outcome
separately
presented
spectrogram
synchro-extracting-transform
obtained
presented.
investigation
with
MobileNet
variants
study
authorizes
better
MobileNetV2
(>99%)
compared
other
schemes
in
study.
Язык: Английский
Optimization of LightGBM for Song Suggestion Based on Users’ Preferences
Journal of Intelligent Systems Theory and Applications,
Год журнала:
2024,
Номер
7(2), С. 56 - 65
Опубликована: Сен. 24, 2024
Undoubtedly,
music
possesses
the
transformative
ability
to
instantly
influence
an
individual's
mood.
In
era
of
incessant
flow
substantial
data,
novel
compositions
surface
on
hourly
basis.
It
is
impossible
know
for
individual
whether
he/she
will
like
song
or
not
before
listening.
Moreover,
cannot
keep
up
with
this
flow.
However,
help
Machine
Learning
(ML)
techniques,
process
can
be
eased.
study,
a
dataset
presented,
and
suggestion
problem
was
treated
as
binary
classification
problem.
Unlike
other
datasets,
presented
solely
based
users'
preferences,
indicating
likeness
specified
by
user.
The
LightGBM
algorithm,
along
two
ML
algorithms,
Extra
Tree
Random
Forest,
selected
comparison.
These
algorithms
were
optimized
using
three
swarm-based
optimization
algorithms:
Grey
Wolf,
Whale,
Particle
Swarm
optimizers.
Results
indicated
that
attributes
new
effectively
discriminated
songs.
Furthermore,
algorithm
demonstrated
superior
performance
compared
employed
in
study.
Язык: Английский