IEEE Women in Engineering Magazine - December 2022 - 20

Still, she was able to flex her creative
muscle and, in some ways, feed her passion
for dramatic storytelling when she
designed and coded an ahead-of-its-time
novel neural network for auto insurance
fraud detection that performed a narrative
analysis of completed insurance
applications as if they were sourced literary
texts.
" The neural network looked through
answers on your application. There
are lifestyle questions about
how many kids, et cetera,
and I used that as the
story someone was
telling about their
lifestyle. I figured
people who already
have a junked car
and are just submitting
now so in a few
months they can submit
a claim-they don't care
what they are filling out. They're
just randomly answering lifestyle questions.
They're not going to fall into a
category of lifestyles. They're going to
have a two-seater sports car and four
kids. Maybe not so blatant, but that
make and model car is rarely bought
with those lifestyle questions, " she says.
As she explains it, she devised a
neural network that " identified literary
genre and flagged all works of fiction for
further investigation. "
" None of that got in the official proposal
of what I planned to do or the final
report on what it did. That was conventional
math, statistics, linear regression
models. My boss was cautious about saying
it was a neural network and not just
a statistical theory that neural networks
are based on, " she explains.
" The thing was, it worked, " she says.
She developed neural networks to identify
the clusters and groups and to flag
those that didn't make realistic sense.
She tested with old fraudulent test
cases, and her work correctly identified
all of them. As she says, some insurance
applicants " just thumbed their nose at
20
the questions " and didn't answer them
seriously, but still, those nonfraudulent
cases could be easily identified while still
saving much clerical time by not having
to review every single application.
It worked so well that when the
Often, she
says, the work
underwriting department tested it they
recommended a production version,
but management was concerned that
it might work on only test data. " They
worried that in the future it might
overfit or underfit the data,
so they never greenlighted
the project, " Weisbecker
says. Still, she greatly
she loves is the work
others grumble about
or say they wouldn't
want to be
stuck doing.
enjoyed the work,
" the only thing she
did at the insurance
company that was
close to her hobby. "
The company was
in financial trouble and
Weisbecker decided to see
it as an opportunity to change
careers. It was time to look into the
entertainment and career potential of
her other hobby, neuromorphic engineering,
which meant it was time to go
back to school.
To ensure she'd be able to graduate
on time, she spent the summer taking
two chemistry courses, with materials
science sandwiched in between. " I like
to refer to it as my summer spent in the
islands, " she jokes. " I spent the summer
on No Fun Atoll. "
What was fun was applying neuromorphic
engineering principles to
design a circuit for her engineering
senior class project. Weisbecker didn't
tell anyone that the three inputs and
8-bit output worked like a simple artificial
retina. She just explained that her
circuit turned on the output bit associated
with the location of the strongest
input signal, even when it was halfway
between two sensors (biologists call
this hyperacuity; it's similar to " halfpixel
accuracy " in computer graphics).
Designing and building that circuit was
as much fun as designing and coding
IEEE WOMEN IN ENGINEERING MAGAZINE DECEMBER 2022
the Speedway video game. And, like
making games that have drama and
suspense, designing circuits for batteryoperated
toys that create the illusion
of being alive would be put on the back
burner until she had a real job.
A few years later, with her engineering
degree in hand, she had choices, and
Lockheed Martin, just starting to hire
again after a freeze, caught her eye. She
thought that a traditional engineering job
would be a good addition to her game consultant
resume. She accepted their offer.
Solving the Difficult, Fun Problems
Weisbecker, like her father, has the inventor's
spirit, wells of creativity, and a
persistent mindset. At home, she says
she is still
looking for ways to turn
her hobbies into commercial products.
These range from the literary (she is
seeking a publisher to create posters of
two of her poems) to the technological:
Weisbecker designs video games where
" breaking the fourth wall " comes from
suspense and the building up and releasing
of dramatic tension during the
natural course of gameplay. She also designs
battery-operated toy " pets " with "
artificial sensory organs " (like her artificial
retina circuit) that create the illusion
of being alive and that interact with
other toys and with people in entertaining
ways.
Throughout the course of her career,
Weisbecker has gravitated toward projects
that are math intensive and that,
like puzzles, require creativity and logic
to solve. She seeks them out, always
looking for an intellectual challenge.
Often, she says, the work she loves is
the work others grumble about or say
they wouldn't want to be stuck doing.
She is happy to say, though, that everywhere
she goes " there are people who
are happy to let me solve the difficult,
fun problems for them. "
-Katianne Williams
IEEE Women in Engineering Magazine - December 2022

Table of Contents for the Digital Edition of IEEE Women in Engineering Magazine - December 2022
