Formulation
of
the
problem.Winter
wheat
yield
and
grain
quality
largely
depend
on
providing
plants
with
mineral
nutrients
throughout
growing
season.Intensive
varieties
are
characterized
by
higher
requirements
for
nutritional
conditions
only
a
full
balanced
supply
can
form
high
yields.Sufficient
in
an
easily
accessible
soil
almost
non-existent,
so
to
get
winter
wheat,
you
need
apply
fertilizers
[1,
2].Analysis
recent
research
publications.Winter
is
very
demanding
conditions.It
since
its
root
system
low
ability
absorb
from
hard-to-dissolve
compounds
soil.The
absorption
determined
primarily
[3,
6].Before
earing
phase,
depending
nutrition
conditions,
absorbs
70-82%
nitrogen,
75-85%
phosphorus
maximum
amount
[2,
5].A
number
authors
believe
that
realization
high-yielding
intensive
technology
it
necessary
90-120
kg/ha
potassium
2].Presenting
main
material.During
2019-2021
field
experiments
were
conducted
study
impact
Zimoyarka
dark
gray
podzolic
soils.Characteristics
arable
layer
forest
as
follows:
pH
-5,7
-6,0,
content
hydrolyzed
nitrogen
(according
I.V.
Tyurin
M.M.
Kononova)
117
mg,
mobile
115
exchangeable
Kirsanov)
129
mg
per
1
kg
humus
Turin)
these
soils
2.2%.Ammonium
nitrate
(GOST
2-85),
granular
superphosphate
-5956-78)
chloride
4568-95)
used
experiments.Phosphorus
part
applied
autumn
pre-sowing
cultivation,
rest
dose
N
30
60
during
fertilization.Results
seeds
given
table
1.
ACM Transactions on Management Information Systems,
Journal Year:
2025,
Volume and Issue:
16(1), P. 1 - 11
Published: Feb. 7, 2025
Large
language
models
have
been
advancing
very
rapidly
and
are
making
substantial
impacts
on
all
areas
of
business
management.
We
review
the
development
large
their
applications
in
management,
identify
major
issues
challenges
faced
by
both
practitioners
researchers.
Based
our
review,
we
propose
an
agenda
for
information
systems
researchers
discuss
some
potential
directions
future
research.
Lastly,
present
articles
special
issue
as
exemplary
research
implications.
Research Square (Research Square),
Journal Year:
2025,
Volume and Issue:
unknown
Published: March 26, 2025
Abstract
Financial
predictive
modeling
plays
a
crucial
role
in
decision-making,
risk
management,
and
strategic
planning
within
financial
markets
institutions.
Ensuring
the
veracity
accuracy
of
synthetic
data
is
major
challenge
when
it
comes
to
developing
forecasting
models.
Otherwise,
inaccurate
model
predictions
flawed
decisions
are
likely
result
if
artificial
created
does
not
look
like
real-world
patterns.
A
research
study
has
tended
apply
generative
techniques
on
information
determine
potential
for
influencing
models
through
content
selection
improved
accuracy.
This
critically
examines
RBMs
Generative
Adversarial
Networks
(GANs)
Variational
Autoencoders
(VAEs)
create
that
mimics
intricate
behaviors
datasets
market
volatility,
price
couplings,
time
lags
perfection.
Furthermore,
this
introduces
use
Kullback-Leibler
Divergence
(KL-Divergence)
as
measure
evaluate
how
distant
from
real
data.
The
operative
nature
KL-Divergence
allows
one
ascertain
well
can
emulate
true
underpinning
distribution
actual
finance
Results
indicate
Real
Fake
achieved
skewed
peaking
at
25
with
density
coverage
fluctuating
−
0.50
1.25
using
Python
software.
results
reveal
integration
generated
reporting
by
R.B.M.
other
into
training
substantially
improve
performance,
even
under
conditions
tend
flip-flop
or
show
rarity.
Posted
literature-On
future
dealing
between
advanced
reinforcement
learning
derive
finest
possible
pools
adaptability
forecasting.
Scientific Reports,
Journal Year:
2025,
Volume and Issue:
15(1)
Published: April 4, 2025
The
purpose
of
this
work
is
to
explore
methods
visual
communication
based
on
generative
artificial
intelligence
in
the
context
new
media.
This
proposes
an
image
automatic
generation
and
recognition
model
that
integrates
Conditional
Generative
Adversarial
Network
(CGAN)
with
Transformer
algorithm.
generator
component
takes
noise
vectors
conditional
variables
as
inputs.
Subsequently,
a
module
incorporated,
where
multi-head
self-attention
mechanism
enables
establish
complex
relationships
among
different
data
points.
further
refined
through
linear
transformations
activation
functions
enhance
feature
representations.
Ultimately,
captures
long-range
dependencies
within
images,
facilitating
high-quality
images
meet
specific
conditions.
model's
performance
assessed,
findings
show
accuracy
proposed
reaches
95.69%,
exceeding
baseline
algorithm
by
more
than
4%.
Additionally,
Peak
Signal-to-Noise
Ratio
33dB,
Structural
Similarity
Index
0.83,
indicating
higher
quality
accuracy.
Therefore,
achieves
high
prediction
generated
quality,
promising
significant
application
value
media
era.
AI,
Journal Year:
2025,
Volume and Issue:
6(5), P. 95 - 95
Published: May 2, 2025
Background:
Investment
decisions
in
stocks
are
one
of
the
most
complex
tasks
due
to
uncertainty
which
will
increase
or
decrease
their
values.
A
diversified
portfolio
statistically
reduces
risk;
however,
stock
choice
still
substantially
influences
profitability.
Methods:
This
work
proposes
a
methodology
automate
investment
decision
recommendations
with
clear
explanations.
It
utilizes
generative
AI,
guided
by
prompt
engineering,
interpret
price
predictions
derived
from
neural
networks.
The
also
includes
Artificial
Intelligence
Trust,
Risk,
and
Security
Management
(AI
TRiSM)
model
provide
robust
security
for
system.
proposed
system
provides
long-term
based
on
financial
fundamentals
companies,
such
as
price-to-earnings
ratio
(PER)
net
margin
profits
over
total
revenue.
explainable
artificial
intelligence
(XAI)
uses
DeepSeek
describing
suggested
well
several
charts
Shapley
additive
explanation
(SHAP)
values
local-interpretable
model-agnostic
explanations
(LIMEs)
showing
feature
importance.
Results:
In
experiments,
we
compared
profitability
portfolios,
ranging
8
28
values,
maximum
expected
increases
4
years
NASDAQ-100
S&P-500,
where
both
bull
bear
markets
were,
respectively,
considered
before
after
custom
duties
international
trade
USA
April
2025.
achieved
an
average
56.62%
while
considering
120
different
recommendations.
Conclusions:
t-Student
test
confirmed
that
difference
index
was
significant.
user
study
revealed
participants
agreed
were
useful
trusting
system,
score
6.14
7-point
Likert
scale.
International Journal of Computer Science and Information Technology,
Journal Year:
2024,
Volume and Issue:
2(3), P. 1 - 9
Published: May 28, 2024
Intelligent
manufacturing
has
gradually
become
an
important
development
trend
in
the
industrial
field.
As
artificial
intelligence
technology,
machine
vision
been
widely
used
field
of
automation.
This
paper
discusses
and
application
robot
arm
intelligent
picking
system
based
on
manufacturing.
The
converts
target
into
image
signal
through
acquisition
device,
sends
it
to
special
processing
for
digital
processing.
Then,
performs
various
operations
extract
features
target,
controls
action
equipment
according
discriminating
results.
With
technology
as
core,
realizes
automatic
tasks
improves
production
efficiency
quality.
FinTech,
Journal Year:
2024,
Volume and Issue:
3(3), P. 460 - 478
Published: Sept. 20, 2024
The
integration
of
generative
AI
(GAI)
into
the
financial
sector
has
brought
about
significant
advancements,
offering
new
solutions
for
various
tasks.
This
review
paper
provides
a
comprehensive
examination
recent
trends
and
developments
at
intersection
GAI
finance.
By
utilizing
an
advanced
topic
modeling
method,
BERTopic,
we
systematically
categorize
analyze
existing
research
to
uncover
predominant
themes
emerging
areas
interest.
Our
findings
reveal
transformative
impact
finance-specific
large
language
models
(LLMs),
innovative
use
adversarial
networks
(GANs)
in
synthetic
data
generation,
pressing
necessity
regulatory
framework
govern
finance
sector.
aims
provide
researchers
practitioners
with
structured
overview
current
landscape
finance,
insights
both
opportunities
challenges
presented
by
these
technologies.
Frontiers in Computing and Intelligent Systems,
Journal Year:
2024,
Volume and Issue:
8(1), P. 112 - 115
Published: May 10, 2024
Artificial
intelligence
introduces
a
fresh
research
perspective
to
digital
image
processing.
However,
its
integration
into
the
curriculum
of
colleges
and
universities
for
teaching
processing
remains
scarce.
This
lack
incorporation
results
in
outdated
course
content,
reliance
on
singular
methods,
simplistic
experiments,
consequently
impeding
effective
hindering
development
well-rounded
innovative
individuals.
Digital
technology
expands
horizons
communication
engineering,
facilitating
more
convenient
modes
people.
For
instance,
video
calls
photo
transmissions
diversify
everyday
transcending
constraints
time
space
by
enabling
online
meetings
fostering
enhanced
possibilities.
Despite
these
advancements,
numerous
challenges
methodologies
merit
thorough
exploration.
Therefore,
this
paper
aims
comprehensively
grasp
both
traditional
deep
learning
approaches
processing,
enhancing
practical
project
proficiency
scientific
exploration
capabilities,
thus
serving
as
valuable
reference
similar
endeavors.