arXiv (Cornell University),
Journal Year:
2021,
Volume and Issue:
unknown
Published: Oct. 26, 2021
Unit
testing
is
a
stage
of
where
the
smallest
segment
code
that
can
be
tested
in
isolation
from
rest
system
-
often
class
tested.
tests
are
typically
written
as
executable
code,
format
provided
by
unit
framework
such
pytest
for
Python.
Creating
time
and
effort-intensive
process
with
many
repetitive,
manual
elements.
To
illustrate
how
AI
support
testing,
this
chapter
introduces
concept
search-based
test
generation.
This
technique
frames
selection
input
an
optimization
problem
we
seek
set
cases
meet
some
measurable
goal
tester
unleashes
powerful
metaheuristic
search
algorithms
to
identify
best
possible
within
restricted
timeframe.
two
generate
pytest-formatted
tests,
tuned
towards
coverage
source
statements.
The
concludes
discussing
more
advanced
concepts
gives
pointers
further
reading
artificial
intelligence
developers
testers
when
software.
ITM Web of Conferences,
Journal Year:
2025,
Volume and Issue:
73, P. 01011 - 01011
Published: Jan. 1, 2025
The
volatility
and
diversity
of
financial
markets
make
it
challenging
for
a
single
portfolio
achieve
better
returns,
therefore,
adjustable
portfolios
based
on
the
risk
tolerance
clients
are
highly
demanded.
However,
traditional
strategies
cannot
meet
this
requirement.
Regarding
issue,
paper
combines
Fuzzy
C-means
(FCM)
with
Upper
Confidence
Bound
(UCB)
algorithm,
Genetic
Algorithm
(GA)
optimizing
UCB
parameters
(GA-UCB)
redefining
fitness
GA
(UCB-GA)
to
construct
an
investment
strategy
that
can
be
dynamically
adjusted.
research
methodology
is
as
follows:
assets
grouped
by
FCM,
using
find
best
cluster
among
groups;
UCB,
UCB-GA,
GA-UCB
used
refine
weight
distribution
cluster.
result
shows
cumulative
return
recommended
significantly
higher
than
Sortino
Ratio
improved
1.18,
Maximum
Drawdown
reduced
8%.
In
terms
weights
optimal
cluster;
from
has
highest
approximately
250%
in
algorithms.
largest
at
3.23,
which
1.5
1.63
respectively.
addition,
26%,
1%
lower
UCB-GA
3%
UCB.
Combining
FCM
GA-
improve
stability
adjusting
weight,
leads
ratios.
Software Testing Verification and Reliability,
Journal Year:
2023,
Volume and Issue:
33(4)
Published: May 2, 2023
Abstract
Machine
learning
(ML)
may
enable
effective
automated
test
generation.
We
characterize
emerging
research,
examining
testing
practices,
researcher
goals,
ML
techniques
applied,
evaluation,
and
challenges
in
this
intersection
by
performing.
perform
a
systematic
mapping
study
on
sample
of
124
publications.
generates
input
for
system,
GUI,
unit,
performance,
combinatorial
or
improves
the
performance
existing
generation
methods.
is
also
used
to
generate
verdicts,
property‐based,
expected
output
oracles.
Supervised
learning—often
based
neural
networks—and
reinforcement
Q‐learning—are
common,
some
publications
employ
unsupervised
semi‐supervised
learning.
(Semi‐/Un‐)Supervised
approaches
are
evaluated
using
both
traditional
metrics
ML‐related
(e.g.,
accuracy),
while
often
tied
reward
function.
The
work‐to‐date
shows
great
promise,
but
there
open
regarding
training
data,
retraining,
scalability,
evaluation
complexity,
algorithms
employed—and
how
they
applied—benchmarks,
replicability.
Our
findings
can
serve
as
roadmap
inspiration
researchers
field.
Mutation
testing
can
help
reduce
the
risks
of
releasing
faulty
software.For
such
reason,
it
is
a
desired
practice
for
development
embedded
software
running
in
safetycritical
cyber-physical
systems
(CPS).Unfortunately,
state-ofthe-art
test
data
generation
techniques
mutation
C
and
C++
software,
two
typical
languages
CPS
rely
on
symbolic
execution,
whose
limitations
often
prevent
its
application
(e.g.,
cannot
black-box
components).We
propose
approach
that
leverages
fuzz
testing,
which
has
proved
effective
with
software.Fuzz
automatically
generates
diverse
inputs
exercise
program
branches
varied
number
ways
and,
therefore,
statements
different
states,
thus
maximizing
likelihood
killing
mutants,
our
objective.We
performed
an
empirical
assessment
components
used
satellite
currently
orbit.Our
evaluation
shows
based
kills
significantly
higher
proportion
live
mutants
than
execution
(i.e.,
up
to
additional
47
percentage
points).Further,
when
be
applied,
provides
significant
benefits
41%
killed).Our
study
first
one
comparing
testing;
results
provide
guidance
towards
tools
dedicated
testing.
Proceedings of the Genetic and Evolutionary Computation Conference,
Journal Year:
2022,
Volume and Issue:
unknown, P. 1317 - 1326
Published: July 8, 2022
Multi-objective
test
case
selection
techniques
are
widely
investigated
with
the
goal
of
devising
novel
solutions
to
increase
cost-effectiveness
verification
processes.
When
evaluating
such
approaches
entire
Pareto-frontier
algorithm
needs
be
considered.
To
do
so,
several
quality
indicators
exist.
The
hyper-volume
(HV)
is
one
most
well-known
and
applied
indicator.
However,
in
context
selection,
this
metric
has
certain
limitations.
For
instance,
two
different
fitness
function
combinations
not
comparable
if
used
at
search
algorithm's
objective
level.
Consequently,
researchers
proposed
revisited
HV
(rHV)
compute
rHV,
each
solution
individually
assessed
through
external
utility
functions:
cost
fault
detection
capability
(FDC).
increases
risk
having
dominated
solutions,
which
practice
may
lead
a
decision
maker
(DM)
select
solution.
In
paper
we
assess
whether
rHV
an
appropriate
indicator
multi-objective
algorithms.
empirically
results
between
FDC
DM
instances
hold.
Long
story
short,
2021 36th IEEE/ACM International Conference on Automated Software Engineering (ASE),
Journal Year:
2021,
Volume and Issue:
unknown, P. 368 - 380
Published: Nov. 1, 2021
Model-based
testing
is
a
structured
method
to
test
complex
systems.
Scaling
up
model-based
large
systems
requires
improving
the
efficiency
of
various
steps
involved
in
testcase
generation
and
more
importantly,
test-execution.
One
most
costly
bring
system
known
state,
best
achieved
through
synchronising
sequences.
A
sequence
an
input
that
brings
given
predetermined
state
regardless
system's
initial
state.
Depending
on
structure,
might
be
complete,
i.e.,
all
inputs
are
applicable
at
every
system.
However,
some
partial
this
case
not
usable
Derivation
sequences
from
complete
or
challenging
task.
In
paper,
we
introduce
novel
Q-learning
algorithm
can
derive
with
structures.
The
proposed
faster
process
larger
than
fastest
sequential
derives
Moreover,
also
recent
massively
parallel
Furthermore,
generates
shorter
