►
From YouTube: DevoWorm (2020, Meeting 31): GSoC Final Presentations (DevoLearn), Agent-based Modeling
Description
Attendees: Bradly Alicea, Mayukh Deb, Ujjwal Singh, Maniak Deb, Susan Crawford-Young, Jesse Parent, Tom Portegys, and Richard Gordon.
B
C
Oh,
I'm
I'm
good.
I
just
don't
have
I
haven't
plugged
in
my
camera
because
my
my.
B
All
right,
how
are.
C
A
A
B
So
why
don't
we
get
started?
Welcome
to
the
meeting.
We
have
our
gsoc
final
presentations
today
and
then
we'll
talk
about
some
of
the
things
that
might
come
after
gsoc
with
respect
to
publishing
something
on
this
and
then
a
couple
other
things
after
that.
So
maya
told
me
that
they
had
their
presentations
ready
to
go
so
who
wants
to
start
first.
B
D
E
B
D
Yeah,
okay,
so
then
I'll
start
so
hello,
everyone,
so
g-shock
has
come
to
an
end
after
like
three
long
months,
not
exactly
long,
but
still.
Sometimes
it
was
so
community
warning.
This
is
just
basically
like
a
brief
look
at
the
developments
and
the
deviations
in
the
last
few
months.
So
yeah
first
is
community
bonding
and
phase
one.
So
initially,
I
just
started
out
with
some
simple
stuff:
just
dig
around
and
see
what
people
learn,
and
I
mean
what
diva
worm
was
all
about.
D
I
actually
developed
another
one
of
them.
This
one
could
skeletonize
the
one
and
say
that,
like
the
save
the
positional
data
as
csv
files,
so
this
is
also
up
there.
This
these
are
not
very
important
later
on,
but
they're
like
just
interesting,
and
these
are
just
explorative
work
so
moving
on
so
then
I
started
off
by,
like
I
just
developed
an
lstm
model
which
could
like
predict
the
curvature
of
the
worm's
body
from
the
time
series
data
containing
the
distance
between
the
head
and
the
tail.
D
Basically,
the
curvature
is
inversely
proportional
to
the
distance
right
for
a
given
like
normalized
coordinates
like
for
a
given
normalized
coordinates
like
set.
This
worked
so,
but
this
was
even
like,
like
it
was
not
perfectly
accurate,
and
it
was
also
part
of
like
the
explorations
that
I
did
so
do
moving
on.
D
So
then,
what
I
did
was
principal
component
analysis
on
spatial
data
from
drive
cells,
so
this
basically
helps
us
visualize
like
how
cells
descending
from
the
common
ancestors
had
common
leg
parameters
like
I
did
principle
component
analysis
on
like
four
parameters.
Those
were
like
position
xyz
and
time,
and
then
I
brought
them
down
into
two
dimensions
x
and
y,
so
I
plotted
them
so
far
for
each
lineage,
as
you
can
see
on
the
graph
to
the
right,
can
you
see
the
graph?
Is
it
large
enough?
D
B
Yeah,
you
can
see
like
yeah.
D
Yeah,
I
don't
think
zooming
in
is
possible.
I
will
share
the
slides
after
this
okay.
So
what
it
shows
is
that
basically,
like
the
the
lineages
which
are
common
as
just
like
on
lie
on
like
similar
positions
within
the
principal
component
space,
this
is
what
it
basically
showed
like.
This
is
an
interesting
visualization
that
things
could
be
done
on
this
in
the
future.
D
So
then
pictures
of
from
the
one
mm
data
set
this
like
I
found
it
like
near
the
end
of
phase
one.
So
then
it
started
working
and
my
like
my
prime
goal-
was
not
exactly
this
data
set,
but
to
build
a
pipeline
for
segmenting
other
stuff
later
on.
So
I
actually
you
decide
kind
of
resetting
this
pipeline
later
on
for
another
one
which
I'll
be
showing
soon.
D
D
So
in
this
graph
the
dashed
lines
are
the
predictions
made
by
a
deep
learning
model
or
someone
who
actually
manually
annotated
the
epic
data
set.
So
it's
pretty
close,
I
guess-
and
it
could
give
this
these
kinds
of
results
consistently
on
different
types
of
videos
or
different
types
of
photos.
So
this
worked
out
pretty
well.
So
then,
this
is
another.
D
D
And
to
the
right,
you
can
see
the
predictions
made
by
the
deep
learning
model
like
the
dashed
lines
that
is
like.
If
you
just
consider
the
a
lineage,
then
it's
the
orange
dashed
lines
versus
the
blue
dashed
lines,
they're
pretty
close
so
like,
and
it
can
do
kind
of
like
live
predictions
from
videos
and
stuff
which
we'll
come
to
later
on.
So
moving
on
the
next
thing
that
I
did
was
like,
it
was
kind
of
a
tangent
again.
D
So
what
it
did
was
like
I
made
a
can,
which
could
generate
synthetic
images
of
the
m
drive
like,
but
we
can
like
develop
this
more
later
on.
Maybe
we
can,
we
could
develop
a
conditional
gang.
D
What
it
could
do
is
that
we
would
give
it
populations
and
it
would
give
me
a
synthetic
image
of
an
embryo
which
has
approximately
those
cell
populations
within
that
image,
so
yeah
until
now
it's
this
is
not
a
conditional
again
right
now,
so
it
basically
returns
for
like
images
of
random
populations
within
the
embryo,
so
yeah
moving
on.
D
Moving
on
yeah,
so
then
deep
segmentation
model
to
segment
the
c
elegans
mri.
This
segmentation
model
was
actually
later
using
divorce
to
build
like
an
easy
to
use
segmentation
tool
we'll
come
to
that
soon.
But,
as
you
can
see
like
this,
video
is
like
a
pretty
like
noisy
kind
of
a
video,
and
then
it
could
segment
like
it
pretty
well
like.
I
think
it
got
an
ad
like
an
iou
score
of
above
85
to
90.
I
guess
I
don't
know
the
exact
like
latest
figures,
and
then
it
could
also
extract
the
centroids.
D
So
moving
on
then
phase
three
so
say
hello
to
david
divolar.
Jivolon
is
a
python
library
which
can
help
accelerate
data-driven
research
on
c
elegance
and
by
automating
the
process
of
collecting
metadata.
So
what
it
does
is
basically
it's
kind
of
a
combination
of
all
the
good
stuff
that
I
made
during
phase
one
and
phase
two,
and
we
just
integrated
some
of
those
models.
D
Not
all
to
form
was
very
simple
to
use
kind
of
a
python
library
which
is
already
there
on
pipeline,
like
people
can
easily
install
it
and
use
it
like
to
automate
the
process
of
collecting
metadata
like
like.
I
could
give
this
example
like
people.
One
could
one
day
have
a
few
various
like
data
given
research
or
like
approaches
like
I'll
show
this
gif
to
the
left.
D
D
So
these
are
the
kinds
of
interesting
things
that
you
can
do
and
on
the
right
you
can
see
another
gif,
which
is
basically
like
an
approximate
map
of
the
centroids
of
the
cells
within
a
c
elegans
embryo,
from
a
given
instance
of
its
embryogenesis
processes.
D
D
So
this
is
basically
the
structure
of
how
devolent
is
right.
Now
there
are
three
main
models:
one
is
a
lineage
population
model.
Lineage
population
is
this:
lineage
population
model
is
basically
this
model
which
can
like
take
images
and
videos,
and
it
can
return
like
csv
files
containing
the
populations
of
each
in
of
of
each
frame
like
frame
by
frame.
It
can
return
in
rows.
D
Only
then
is
the
model
file
downloaded,
which
is
like
50
mb,
each
so
yeah
moving
on
then
looking
ahead
with
the
right
contributor
civilian
could
one
day
become
a
useful
tool,
so
these
are
basically
the
things
that
could
be
improved
in
development,
which
I
thought
about
lately.
Some
of
them
are
also
on
the
github
issues.
You
could
check
them
out
so
use
the
contributors
could
help
by
improving
the
model's
accuracies.
D
They
can
submit
more
deep
learning
models
which
could
be
added
if
they
are
in
alignment
with
our
interests
like
on
the
c
elegance
or
embryogenesis
itself,
and
if
they
are
accurate
enough
and
then
development
currently
works
with
only
images
and
video
files
and
csvs,
and
it's
not
like
it,
cannot
take
dot
tip
for
dot
mac
files
like
as
input.
So
we
could
extend
that
support
soon
enough,
like
someone
could
do
that
I'll
have
to
raise
an
issue
on
this
and
then
it's
it
currently
works
on
the
gpu.
It
currently.
B
Questions
so
yeah.
This
is
the
summer
code
project
for
the
pre-trained
model.
So
for
those
of
you
who
don't
know,
we
have
the
diva
learn
platform
which
is
divo,
learned
the
program,
and
then
we
have
usuals
what
he's
going
to
present
on,
which
is
the
revamp
of
devo
zoo
and
then
also
some
other
development.
B
So
this
is
the
link
to
diva
learn
in
the
chat
like
just
put
it
in.
If
you
want
to
the
the
organization,
is
devo,
learn
and
then
there's
a
diva
learn
repository
underneath
that
and
you
can
download
his
program
and
and
work
with
it.
We
also
should
talk
about
the
you
know.
If
you
want
to
do
anything
with
the,
I
don't
know
how
it's
integrated
into
this,
but
the
movement
data
that
you
were
showing
and
what
else
was
there?
So
what's
yeah.
B
So
that's
the
quick
summary
so
so
he's
writing
up
his
g
sock
proposal
so
anyways.
I
encourage
you
to
go,
look
at
it
and
you
know
play
with
it.
If
you
want
and
see
you
know
if
you
want
to
contribute
their
number
of
pathways,
we're
kind
of
setting
up
issues
on
that.
B
B
We
can
set
up
an
issue
board
for
that
yeah
and
you
can
try
out
diva
learning
in
a
google
collab.
So
my
oak
has
made
a
nice
collab
notebook
of
everything
in
diva
learn.
B
So,
let's
move
on
to
and
hear
from
him.
E
B
E
E
Sorry,
so
just
a
quick
good
like
news
that
attracts
me
about
the
neural
link
like
quit
like
if,
like
I
know
most
of
the
people
you
know
but
like
the
main
thing
I
want
to
highlight
about
literally,
is
about
the
size
that
they
are
getting.
They
are
making
it
much
more
compact
and
compact
which
like
amazes
me,
so
I
think
like
it,
should
be
given
a
little
emphasis
on
it,
so
that
we
can
discuss
it
about
later.
E
The
presentation
so
yeah
hi-
I
am
most
of
you-
people
know
me-
I
am
a
second
year
or
third.
E
It
is
given
to
me
this
year
and
I
was
fortunate
enough
to
do
it
so
aim
of
the
openness
is
to
connect
the
teamwork
and
open
ml
architecture
and,
like
I
just
did
to
improve
the
web
interface
and
to
develop
new
intuitive
and
beautiful
interfaces.
Reorganize.
E
Our
devices
we
are
coming
in
few
minutes
on
it
open
level
cell
tool,
customization
new.
E
Is
something
that
has
been
prepared
by
in
last
year,
and
this
year
we
tried
to
improve
it,
and
I
incorporate
much
more
things
in
it
and
like
to
give
a
fresh
look
completely
editor.
So
this
is
the
thing
that
we
have
completed.
I
have
completed,
it
is
not
linked,
but
it
will
be
late
this
week
so-
and
this
is
like
kind
of
latest
researchers
don't
have
to
like
use
and
like
download
and
label.
E
So
I
like
to
have
a
very,
very
little
contribution
to
it
and
like
more
of
like
kind
of
a
tutorial
based
and
like
helping
setting
it
up.
So
it
is
a
very
good
project,
like
I
must
say
so.
Community
bonding
period,
so
community
bonding
period
is
like
one
of
the
phase
where
we
have
interacted
with
many
people
so
like
we
have
bonded
with
people's
like.
We
have
got
opinion
from
many
peoples.
E
We
have
get
like
how
to
proceed
further.
Then
we
have
me
weekly
meeting
and
onboarding.
So
bradley
and
finna
have
onboarded
us
on
the
project.
They
have
told
us
like
what
we
have
to
do
and
how
we
have
to
manage
things
and
all
these
stuffs
in
a
meeting.
And
then
we
have
office
hours
where
we
have
described
our
project
in
detail,
and
we
got
some
few
feedbacks
from
the
community
to
work
on
which
are
very
beneficial
for
us
and
like
we.
C
E
E
E
This
is
basically
about
the
duo,
ai
tool,
and
it
consists,
as
you
can
see,
all
the
history
of
their
warm
like
other
tools
for
quite
getting
started
with
the
neural
biology.
F
E
E
E
E
Of
helping
researchers
around
the
world
so
like
this
website
already
exists.
What
I
have
done
this
year
is
like
to
give
a
new
fresh
look,
so
that
people
don't
have
to
click
on
all
the
links
and
like
have
to
wait
for
like
oh,
what
is
the
latest
it
is.
He
has
to
go
and
sometimes
like
it
is
arriving.
So
people
have
to
ask
for
permission
so
like
they
can
readily
see
on.
E
E
This
year
we
tried
to
incorporate
more
things
to
the
existing
model
and
app
and
to
provide
the
full
videos
of
the
owner
and
see
on
website
itself.
So
it
is
not
just
a
tool.
It
is
like
a
resource
center
where
you
can
come
and
read
about
seed
or
delegates
segmentation
procedure.
You
can
read
about
a
library
that
you
have
built
recently.
You
can
driver
tools
online,
you
can
contact
maintainers,
so
the
kind
of
a
mini
resource
center
you
can
say
under
devops
ai.
E
So
basically,
the
aim
of
ai
is
to
connect
all
the
things
that
they
want.
Group
has
on
one
portal
like
under
one
roof
so
that
all
the
things
can
be
accessed.
You
don't
have
to
go
here
like
you
have
to.
If
you
want
to
access
some
data,
you
have
to
ask
us,
like
you
want
that
data.
We
have
to
provide
information.
E
C
E
As
I
have
said
like,
it
is
majorly
mixed
project,
so
I
have
a
little
bit
contribution
on
it,
so
it.
F
A
E
On
the
older
library
and
page
itself,
like
there
are
many
visitors
on
our
packages
and,
like
all
this,
like.
E
So
before,
like
thanks,
I
would
like
to
just
let
you
see.
E
E
Segmentation
model,
so
we
have
created
some
notebooks
for
people
to
get
on
board
about
what
image
segmentation
is.
So
we
have
tried
to
give
them
as
much
detail
as
we
can
like.
If
a
person
is
interested
in
going
deep,
like
how
the
image
segment
is
done,
and
he
has
some
experience
in
computer
science,
he
can
go
as
as
like
dean.
We
have
made
it
possible
like
as
we
can,
he
can
go
and
see
what
the
things
are
really
how
things
are
really
happening
in
that
notebook.
E
E
E
E
The
main
people
who
are
involved
in
this
group
and
after
that
you
can
contact
us
through
your
like.
If
you
have
any
questions,
you
have
anything
that
you
want
to
contact
and
you
want
to
make
things
right,
so
you
can
definitely
contact
us
if
you
want
to
like
have
have
some
bugs
you
want
to
report
us,
you
can
do
that
as
well.
So
there's
a
kind
of
menu
here
so
moving
forward
c
dot
electrons.
So
there
is
segmentation
tool
that
we
are
talking
about.
E
E
In
the
area
of
itself,
you
can
download
these
things
in
a
form
of
csv
file
from
our
portal,
so
there
is
a
project.
E
Get
up
and
like
don't
have
to
get
around
like
or
label
each
and
every
cell,
it
is
in
an
unsupervised
manner,
so
it
will
make
research
quite
fast.
So,
as
you
can
see,
this
is
the
intro
to
the
introduction
about.
E
Process,
so
this
is
the
thing
that
we
have
taken
from
atlas
which
is
cited
below
so,
as
you
can
see
like.
These
are
the
different
phases
like
fertilization,
poly,
fresh
gastrulation,
elongation,
etc.
So
you
can
visit
format
list
or
to
know
more
about
it,
and
it.
E
Divider
library
link
where
you
can
get
quickly
started
about
like
how
to
install
they
will
learn.
What
are
the
different
functions
that
you
can
use
and
all
these
things,
so
it
has
as
simple
as
you
can,
so
you
can
just
take
a
few
commands
and
you
can
do
your
thing.
You
don't
have
to
just
go
deep
into
like
what
are
the
tools
and
what
are
the
functions
that
are
being
responsible
for
producing
results?
You
just
have
to
use
the
library
or
the
abstract
level
so
that
you
just
know
the
coins.
E
E
E
So
I
don't
have
cds
photo
right
now
in
my
this
laptop
so,
but
you
can
definitely
upload
your
photo,
then
upload
it.
It
will
produce
a
result.
I
have
tested
it
multiple
times,
then
the
contact
information
like
if
you
have
any
issues
or
any
trouble
like
if
you
are
facing
you
think
like
something
has
to
be
improved.
Something.
E
E
E
E
Website
so
we
have
not
removed
anything
from
here
as
of
now.
Oh,
I
guess
I
have
renamed
it,
but
nevertheless
I
remember
so
yeah.
This
is
the
older
people
which,
as
you
can
see
like
it,
has
all
the
things
there
right
there.
It
has
nothing
to
be
like
which
is
which
are
extra
there,
but,
like
you,
have
to
click
on
certain
links
to
get
more
about
like
what
kind
of
data
you
are
providing.
Are
you
providing
microscopy
images?
Are
you
trying
to
provide
like.
E
E
E
Set
and
all
these
things
so
after
all
these
data
sets,
this
is
the
latest
tweets
like
or
like
it
is
like
kind
of
a
extra
function
like
you
can
get
to
know
like
this
orthogonal
lab
and
open
home
existing
twitter.
So
if
you
are
very
interested,
you
can
follow
us
there
as
well
right
right
from.
E
We
work
on
a
way
we
are
working
right
now
as
well.
It
will
get
a
very
useful
tool,
at
least
I
believe
it.
Then
it
is
basically
like
our
vision
and
research
summary.
There
are
some
microscopic
tutorials
like
for
the
people
who
are
new
in
biologic
field
and
microscopy
field.
They
can
get
some
very
cool
microscopy
tutorials
from
here,
so
this
is
about
like
so
this
is
the
like.
Just
of
the
work
which
we
have
done.
E
E
E
E
So
that's
the
this
year
journey.
I
have
a
wonderful
journey
with
all
the
people
here
like
we
got,
things
are
very
really
smooth
here,
like
I
even
got
sick
for
about
eight
to
nine
days
between,
but
still
we
are
able
to
complete
a
lot
of
things.
B
Yes,
oh
well,
thank
you,
joel,
that's
good!
Any
questions.
B
I
had
a
couple
of
comments,
so
the
one
the
picture
you
showed,
the
uml
diagram.
B
E
E
B
E
Put
it
into
a
contributor
contribution-
and
I
have
also
put
it
on
the
document
that
I
am
preparing
for
the
new
contributors,
which
it
is
almost
complete
like
as
I'm
trying
to
make
it
as
it
did
as
I
can
so
that
contributor
has
never
like
even
have
to
reach
me
for
getting
something
again,
everything
there.
If
we
like
read
it
so
I'm
trying
to
make
it-
and
I
will
be
like
yeah-
definitely
posting
it
there
as
well
yeah.
B
B
That's
good,
okay,
thank
you
and
then
I
have
a
question
from
susan
and
while
tom
had
to
go,
he
said
super
presentations.
B
Susan
says,
would
diva
learn,
be
good
for
high
school
students,
high
school.
E
Learners
like
yeah,
I
think,
like
I
don't
know
like
how
advanced
like
biology,
is
being
taught
to
like
high
school
students
like
because
I
believe
me
and
mike
have
like
dropped
biology
after
like
like
10th
standard,
we
have,
we
don't
have
biology
like
since
last
four
five
years,
so
I
don't
know
like
what
kind
of
curriculum
is
there
for
biology.
F
E
B
Yeah,
I
mean
well,
I
guess
it's
yeah,
so
I
I
could.
Let's
talk
science
know
about
this
site
yeah.
So
let's
talk
science,
I
guess
as
a
teaching
resource
isn't
in
yeah,
so
yeah.
Definitely,
if
you
know
of
any
groups
that
might
be
interested,
you
should
make
this
public
now
it's
open
and
if
you
need
any
links
you
can
contact
me
or
bojolo
or
mayok
about
that
and
just
you
know
we
can
give
you
the
proper
links
and
everything
make
sure
that
you
know
we
get
the
word
out
there.
So
yeah.
B
I
definitely
think,
though,
from
my
perspective,
I
think
high
school
students
could
probably
use
this.
I
definitely
think
if
they
have
and
a
lot
of
high
school
students
nowadays
have
a
programming
background
of
some
type.
So
it's
you
know
something
they
could
probably
do
again.
If
you're
like
ap
biology
student,
maybe
you
could
probably
do
it,
maybe
even
a
you
know
a
regular
biology
course.
B
Maybe
upper
level
10th
or
11th
or
11th
or
12th
grade.
You
know
you
could
do
it
so
yeah
I
mean
I
think,
that's
that's
worth
exploring.
I
bet
you
there's
a
untapped
market
there
and,
let's
see
my
knock
says
his
presentations
are
amazing.
Thank
you
and
my
oak
says
I
have
another
thing
to
say
regarding
divalern,
so
what
was
the
other
thing.
B
D
D
So
it's
a
it's
something
in
which
everyone
actively
like
everyone
in
the
open
source
community
activity
participate
because
they
get
like
free,
goodies
and
free
t-shirts
and
stuff.
So
I
guess
this
year
before
one
should
participate
in
hector
fest.
I
could
arrange
for
that.
So
if
that
happens
then
like
potential
like
potential
contributors
could
bump
into
this
organization
and
then
start
country
contributing
they
might
started
with.
D
E
Like
participating
as
a
developer,
a
page
would
be
very
nice
like
we
do
have
different
kind
of
projects
there
like
we
have
python
notebooks
project.
We
have
a
library
project,
we
have
a
hero
and
python
project
so,
like
people
might
be
interested
in
like,
and
it
is
like,
as
we
all
know,
like
most
other
people
who
come
to
hacktoberfest
just
made
one
or
two.
D
D
B
I
definitely
think
that
would
be
a
good
thing
to
target
I
had.
I
think
we
had
did
this
last
year
with
like
the
devozu.
B
B
E
C
B
Had
yeah
just
he
says
he
can
also
help
with
advertising
that
so
yeah
that'll
be
something
too
we
can
tie
into
to
advertise.
This
jesse
says
these
presentations
are
great
and
I
can't
wait
to
start
building
our
more
things
with
them.
So
yeah.
I
think
that's
good.
I
think
definitely
we
should
be
building
on
this
yeah.
We
should
be
building
on
this
and
you
know
trying
to
use
these
things
like
hacktoberfest
and
other
things
to
push
it
forward
and
get
people
involved
so
well.
Thank
you.
B
Mayor
can
usual
for
those
presentations
and
I'll
remind
you,
and
I
reminded
you
in
slack
yesterday
that
the
well,
I
guess
you
have
your
presentations
or
your
your
submissions
submitted.
So
I
don't
have
to
remind
you
of
that.
So
yes,
yes,
yeah,
okay,
so
good
cool
yeah!
So
that's
I
guess!
That's
it
on
your
end
for
gsoc!
B
Now
it's
my
turn
to
do
your
evaluations
and
again
it's
like
you
know,
like
the
other
ones,
and
so
then
that'll
be
done
and
then
we'll
have
this
project,
and
you
know
people
can
make
sure
that
they
advertise
it
and
definitely
my
usually
you
should
put
this
on
your.
You
know
your
resume
or
your
cv
and
make
it
somewhat
prominent.
B
We
were
talking
about
this
in
my
other
meeting
on
saturday
that
you
know
putting
together
a
good
website
or
presentation
of
your
work
is,
is
good,
I
think
essential,
sometimes
for
people
to
understand.
What's
going
on,
so
you
know
you
should
have
like
a
line
on
your
resume.
It
says
you
know,
diva,
learn
and
you
know
what
what
you
did
in
that
process
and
then
have
some
graphic
or
visualization
associated
with
it.
So
that's
that's.
Definitely
something
that
you
should.
You
know
play.
F
B
It's
like
a
you
know
something
for
your
portfolio.
I
think
you
both
did
a
great
job
there's
another
thing
too.
I
wanted
to
talk
about
which
was
like
papers
that
we
might
derive
out
of
this
work.
So
let
me
share
my
screen
and
I
know
we
talked
about
some
of
this
before.
B
So
let
me
go
into
this,
we
have
the
so
I
think
I
talked
to
mayak
about
this
about
submitting
to
the
journal
of
open
source
science,
and
so
this
is
the
it's
a
journal
where
you
just
it's
a
very
simple
format.
You
describe
a
piece
of
software
open
source
software.
B
You
have
a
couple
of
links
here.
You
have
a
repository,
an
archive
review
and
then
we
have
authors-
and
we
have
this
summary.
So
this
is
a
grid
app
an
extensible
finance
element
toolbox
in
julia.
This
is
the
summary
of
it.
So
there's
a
summary
and
then
there's
some
a
piece
of
code
that
they
put
in
just
as
a
demo
and
then
they
kind
of
describe
why
it's
important.
B
So
this
one
doesn't
have
subsections,
but
they
basically
go
through
the
software.
Why
it's
important
how
it
works?
The
feat
major
features
of
the
software,
some
motivation
of
it,
which
is
you
know,
sometimes
it's
there's
a
clear
motivation.
Sometimes
we
just
did
it
because
it's
a
need
in
the
community.
B
So
in
this
case
they
had
a
really
strong
need.
You
know
really
strong
motivation
for
it,
and
some
of
these
papers
are
more
like
the
software
is
more
academic
oriented,
so
there's
a
strong,
motivate,
stronger
motivation
than
their
otherwise
would
be,
and
so
that's
basically
the
whole
paper.
It's
like
two
or
three
pages,
and
then
you
put
the
references
at
the
end.
So
it's
something
that
we
can
do
pretty
quickly,
but
I
think
it's
important
for
another
way
to
get
things
out
there.
B
So
my
oak
sent
me
a
draft
and
I
sent
him
comments
and
he
answered
the
comments.
So
here's
where
we
are
with
this,
we
have
the
summary
we
have
a
statement
of
need
which
is
good,
and
then
we
have
to
work
on
this
a
bit
more
to
flesh
out.
Maybe
you
know,
has
to
be
a
little
bit
longer
on
this.
I
think,
but
we
can
flush
it
out
a
bit
I'll
put
some
more
work
into
this
and
then
turn
it
back
around
my
hook
and
then
at
the
submission.
B
You
know
you
just
go
to
their
actually,
they
run
it
all
over
the
github
repo,
so
you
know
we'll
submit
it
in
in
tax.
I
think
that's
how
they
submit
their
paper,
so
we
have
to
convert
it
from.
You
know
it's
easier
to
work
in
in
google
docs
for
collaboration,
but
we
can
create,
I
think,
there's
a
text
template
that
they
use.
We
can
get
that
and
put
everything
in
and
then
submit
it
and
so
that'll.
You
know
that'll
be
a
pretty
easy
shot
for
that
and
then
there's
another
opportunity
here.
B
This
is
a
journal
called
patterns,
and
so
this
is
a
bit
more.
Is
these
papers
are
a
bit
longer
and
again
it's
you
know
they
they're.
Basically
discussions
of
data
science
or
some
tool
that
you
build,
and
so
this
might
be
a
place
to
put
everything
together
like
all
of
the
diva
learn.
B
Evil
learn
is
like
a
unified
thing
like
so
this
is
not
just
this
little
piece
of
software,
but
this
is
the
like
the
entire
platform
as
it
were,
and
so
this
would
include,
like
you
know,
mayak's
work
and
as
well
as
work
and
vinay's
work
from
you
know
last
year,
and
you
know
we
can
build
it
up
into
this
unified
platform.
B
Now
I
don't
know
how
I
guess
it's
just
this.
This
is
this
paper
is
on
deep
learning,
reproducibility
for
deep
learning,
so
they
kind
of
go
into
an
introduction.
They
talk
about
some
of
the
concepts
and
terminology
with
respect
to
reproducibility
how
deep
learning
works.
So
it's
at
a
little
bit
higher
level
or
a
little
bit.
You
know
more
detail
than
the
then
the
joss
paper,
but
this
is
still
something
that
is,
I
think,
generally
doable
for
something
like
this.
I
think
it's
it's
self-pressed,
so
it's
a
pretty
fairly
prestigious
journal.
B
F
B
Go
we
can
do
other
things
with
it.
It's
not
a
not
a
total
loss,
but
I
mean
last
year
we
did.
We
did
the
manuscript
on
the
basilaria,
and
so
that
became
a
book
chapter
pretty
quickly
after
it
was
written.
We
wrote
it
over.
B
We
did
the
stuff
last
summer,
so
that
was
vinay
and
usual
and
asmit
and
a
couple
of
other
people
in
the
group,
and
we
submitted
that
to
the
we
put
that
on
the
bio
archive
in
I
think
december,
and
then
we
got
the
book
chapter
accepted
in
maybe
like
april
or
may
so
it
was.
You
know,
pretty
quick
turnaround
several
months,
but
that's
basically
you
know-
and
I
did
this
with
my
other
group
last
summer
too.
B
B
So,
let's
see
maybe
we'll
have
time
for
a
paper
or
two,
I
think
yeah.
I
think,
if
they're
no
more
well,
we
can
ask
questions
about.
You
know
this
offline.
I,
like
the
hacktoberfest
idea
and
we'll
keep
thinking
about
ways
to
make
this
public.
I
think
some
of
you
have
seen
some
of
the
tweets
that
I
sent
already
about
divalearns,
so
we'll
just
try
to
keep
up
that
momentum
of
the
blog
post
as
well,
so
we'll
try
to
keep
the
momentum
up
on
that.
B
We
have
a
couple
more
messages:
okay,
good
we've
corrected
the
c,
elegans,
spelling
and
then
jesse
says
I
can
also
help
with
the
advertisement
very
good.
So
I
think
it
was
last
meeting
in
the
meeting
before
we
talked
a
little
bit
about
agent-based,
modeling
and
diffusion-limited
aggregation
and
all
this
stuff.
Actually,
my
other
group
did.
We
were
also
doing
some
things
in
agent-based
modeling.
So
this
has
been
a
sort
of
the
top
of
my
head,
and
I
think
I
put
this
paper
in
the
presentation
for
the
other
group.
B
Agent-Based
modeling,
so
agent-based
modeling
is
something
that
is,
you
probably
know
what
it
is,
but
it's
not
usually
talked
about
with
that
name.
So
agent-based
modeling
are
things
like
cellular
automata.
B
The
diffusion,
limited
aggregation
model
that
we
saw
a
couple
weeks
ago
is
a
agent-based
model
and
there
are
other
examples
using
that
platform
that
I
showed
you
netlogo
and
so
some
of
the
advantages
and
challenges.
This
is
a
review
of
this
agent-based
modeling
applied
to
morphogenetic
systems
and
then
how
it's
used
and
how
it
can
be
used.
B
So
the
abstract
is
the
complexity
of
morphogenesis
poses
a
fundamental
challenge
to
understanding
the
mechanisms
governing
the
formation
of
biological
patterns
and
structures.
B
So,
in
particular,
the
theoretical
concepts
of
reaction,
diffusion
systems
and
positional
information
proposed
by
alan
turing
and
lewis,
wolpert,
and
so
alan
turing
and
as
you'll
see.
If
you
go
to
our
our
divo
diva
worm
curriculum.
B
We
have
a
bid
on
a
little
tutorial
on
reaction,
diffusion
systems
and
alan
turing's
approach
to
it,
the
chemical
morphogenesis
and
then
lewis
wolpert,
which
is
the
positional
information,
which
is
where
you
know
the
idea
that
cells
use
their
position
to
sort
of
signal
to
each
other.
You
know
what
to
become
or
their
role
in
the
embryo
and
so.
B
Okay,
development
is
not
a
reaction
to
fusion
system.
Yeah,
I
mean
it's
not
like.
Turing
proposes
chemical
morphogenesis,
so
they
yeah.
This
is,
I
don't
know
what
yeah
I
haven't
gone
through
the
I
don't
know
exactly
what
they're
referring
to
in
that
sense,
but
dramatically
influenced
their
general
view
morphogenesis,
although
typically
in
isolation
from
one
another
in
recent
years,
engine
based
modeling
has
been
emerging
has
been
emerging
as
a
consolidation
and
implementation
of
the
two
theories
within
a
symbol.
B
Single
framework
agent-based
models
are
unique
in
their
ability
to
integrate
combinations,
heterogeneous
processes
and
investigate
the
respective
dynamics,
especially
in
the
context
of
spatial
phenomena.
In
this
review,
we
highlight
the
benefits
and
technical
challenges
associated
with
abms
as
tools
for
morphogenetic
events
or
examining
morphogenetic
events.
B
These
models
display
unparalleled
flexibility
for
studying
various
morphogenetic
phenomena
at
multiple
levels
and
have
the
important
advantage
of
informing
future
experimental
work,
including
the
targeted
engineering
of
tissues
and
organs.
So,
okay,
so,
let's
see.
B
B
Then,
a
decade
later,
alan
turing
proposed
in
his
treatise
the
chemical
basis
of
morphogenesis,
a
mechanistic
explanation
that
dominated
the
field
for
several
decades,
and
so
the
core
concept
of
this
theoretical
explanation
was
the
by
now
widely
accepted
rd
mechanism
which,
under
the
right
conditions,
a
two-molecule
reaction
system
is
capable
of
producing
periodic
patterning
through
diffusion
instability
and
says,
is
a
little
bit
yeah.
This
isn't
like
there.
You
know
different
ways.
You
can
look
at
the
reaction.
Diffusion
model
and
people
have
used
this
in
terms
of
describing
patterning.
B
So
this
is
like
a
very
limited
concept,
usually
when
you
look
at
the
way
that
these
chemical
morphogenetic
systems
are
implemented,
they're,
usually
trying
to
predict
patterns
like
striping
patterns
or
like
the
coat
of
a
animal
where
they
have
spots,
or
you
know
the
wing
of
a
an
insect
where
they
have.
You
know
different
coloration
patterns,
and
so
the
cells
are
actually,
you
know
achieving
some
sort
of
order
through
this
process,
and
so
that's
what
they're
getting
at
with
this.
B
And
so
then
the
rd
patterns
produced
by
the
inhibitor
and
activator
gradients
can
be
considered.
Chemical
pre-patterns
that
act
as
templates
for
future
differentiation,
and
so
the
initial,
the
apparent
initial
homogeneity
of
an
egg
or
a
cell
cluster
were
sent
to
spatially
distinct
profiles
of
an
invisible
set
of
high
and
low
concentration
regions.
B
F
B
So
agent-based
models
are
a
theoretical
model,
much
more
even
than
chemical
morphogenesis,
which
is
really
a
theoretical
sort
of
description
of
what's
going
on
in
development.
Is
it
the
best
mo?
Is
it
the
best
theory?
Well,
probably
not
because
it's
it's
not,
you
know
it's.
Maybe
it's
limited
in
in
terms
of
its
scope,
but
we
can
use
it
as
a
use
is
something
that
might
be
useful
to
describe
some
of
this
stuff.
That's
going
on
and
then
so
lewis
wolpert
talked
about
positional
information
as
a
mechanism
of
pattern
formation
during
morphogenesis.
B
B
So
this
this
was
inspired
by
old
observations
during
the
morphogen
genesis
of
sea
urchins,
which
hondrite
hans
dreich
had
made
as
early
as
1891.,
so
in
in
positional
information,
a
cell
was
able
to
determine
its
assigned
fate
from
its
position
relative
to
other
parts
of
the
organism.
The
position
in
turn
is
characterized
by
the
concentration
of
a
morphogen
then,
which
is
a
theoretical
thing.
It
could
be
any
sort
of
like
signaling
molecule,
thus
the
cell
senses
its
positional
value
by
interpreting
a
morphogen
concentration
and
makes
a
fate
fate
decision
based
on
this
local
information.
B
So
it's
different
from
the
reaction
diffusion
in
that
the
spatial
location
is
very
specific.
It
with
reaction.
The
reaction
diffusion
model
is
more
general
in
terms
of
like
gradients
and
things
like
that,
but
this
actually
has
there
is
no
pre-pattern
in
the
embryo
in
terms
of
posit
so
positional
information
makes
this
claim
that
there's
no
pre-pattern
in
the
embryo
and
so
a
reasonable
biological
implementation
of
positional
information
could
be
a
morphogen
source
leading
to
a
spatial
morphogen
gradient
that
gradually
decreases
with
the
distance
from
the
source.
B
This
is
also
very
useful
and
so
as
well,
but
there
are
some
caveats
so,
as
wilpert
himself
recently
stated,
there
is
no
good
evidence
for
for
the
quantitative
aspects
of
any
of
the
proposed
gradients
and
details
of
how
they
are
set
up.
So
it's
not
really
clear
how
these
models
are
implemented
in
living
organisms.
It's
just
like
a
theory
that
maybe
is
very
broadly
fits
the
data,
but
it's
not
you
know.
Sometimes
it
does
seem
a
bit
far-fetched,
but
nevertheless
we
can
reduce
this
to
mathematics.
B
B
Can
we
basically
solve
the
equations
of
this
computational
model,
and
so
that's
how
your
agent-based
models
come
in
where
they
come
in,
because
you
have
the
ability
to
simulate
highly
parallel
system
using
this
basic
theoretical
framework,
an
equation
you
know
which
might
take
a
lot
of
time
to
solve,
or
it
might
be
hard
to
like
generalize
to
like
a
large
population
of
cells,
since
you
may
be
describing
maybe
like
two
cells
that
are
coupled
or
something
like
that,
a
very
general
relationship
or
a
very
specific
relationship
with
react.
B
Where,
with
your
agent-based
models,
you
can
do
this
in
parallel
and
you
can
simulate
populations,
and
you
can
do
this
simultaneously.
So
you
can
get
these
things.
You
can
study
things
like
emergence
and
you
can
use
a
set
of
parameters
to
sort
of
describe
what's
going
on
in
that
population.
B
B
So
you
can
use
things
like
a
lattice
which
are
a
grid
of
cells
where
there's
like
a
single
cell
in
the
middle
and
then
it's
updating
based
on
what
the
state
of
its
neighbors,
you
have
the
cellular
pots
model,
which
is
actually
something
that
we've
discussed
with
respect
to
copy
cell
3d,
which
is
a
platform
that
you
know.
We've
tried
to
get
working
in
our
group,
but
we
haven't
really
gotten
very
far
on
it.
B
The
lattice
is
more
representative
of
a
of
a
cellular
automata,
so
we've
done
things
with
more
morphozoic,
for
example,
that's
a
lattice
model
and
then
their
lattice
free
models
which
are
actually
more
where
they.
You
know
you
take
a
single
centroid
and
you
calculate
a
radius
and
you
calculate
some
sort
of
sphere
and
you
use
that
as
a
way
to
like
you
know,
delimit
you
know
where
the
cell
is
being
influenced
by
so
everything
around
the
cell.
You
know
there
may
be
a
lot
of
signals
going
on
around
the
cell.
B
The
cell
has
maybe
like
a
radius
of
influence
and
within
you
know,
whatever
is
within
that
radius
is
influencing
the
cell.
So
it's
not
like.
It's
not
limited
to
just
grid
squares
or
the
sort
of
these
sort
of
grid
points.
You
actually
have
a
radius
that
can
be
expanded
or
contracted,
and
that
can
lead
to
different
results.
So
you
you
have
different
ways.
You
can
simulate
these
systems.
B
And
you
can
do
different
things.
You
can
set
up
different
types
of
theoretical
scenarios,
so
you
can
use
things
like
a
proliferation
model,
a
migration
model
differentiation
model
to
describe
sort
of
morphogenesis
you
can
use.
You
know
you
can
program
this
in
different
ways
so
that
you
have
different
things
that
are
influencing
each
cell.
B
B
B
And
so
the
one
of
the
key
features
of
agent-based
models
is
that
they
can.
They
specify
unique
growth
rates
for
each
cell
type
within
a
simulation,
so
you're
dealing
with
cells
that
proliferate
and
that's
really
one
of
the
key
morphogenetic
factors
in
an
agent-based
model,
and
so
in
a
developmental
context.
B
So
this
idea
of
proliferating,
cells
and
producing
cells,
different
populations
of
cells
is
key
to
you
know
how
an
abm
generates,
complexity
and
emerging
complexity
in
particular,
and
so
there
are
different
ways
that
you
can
observe
things
like
patterning.
So
in
a
aging
based
model,
you
know
you
observe
patterning
and
things
like
this,
but
you
have
to
figure
out
ways
to
make
that
happen,
and
it's
not
really
magic.
It's
just.
B
You
have
to
figure
out
which
types
of
rules
you
need
to
implement
to
get
it
to
sort
of
mimic
what's
going
on
in
nature,
so
it's
a
bit
harder
than
writing
an
equation,
although
maybe
it
isn't
because
in
an
equation
you
have
to
specify
different
variables
and
parameters,
and
you
know
relationships
and
you
have
to
you
know,
sort
of
hypothesize
those
things
and
you
know
it.
But
it's
harder
to
you
know
it's
harder
to
really
kind
of
see
the
outcome
right
away
with
an
agent-based
model.
B
But
it's
a
good.
It's
a
good
model,
it's
a
good
hypothesis,
and
so
let's
see,
and
so
they
yeah
they
really
get
into
a
lot
of
detail
here
and
if
you're
interested
in
agent-based
models.
This
is
a
good
paper
to
go
into
in
depth.
B
B
B
You
can
use
the
different
ways
of
modeling
interactions
between
cells
and
build
these
different
structures
that
allow
you
to
model
cells,
and
then
the
cells
run
in
the
simulation,
and
then
they
produce
outcomes
that
maybe
they
look
like
something
that
you
would
see
in
nature,
and
so
you
know
there
are
a
bunch
of
different
ways
to
do
this.
They
introduce
three
ways
to
do
this.
B
What
else?
So,
then?
They
also
have
abms
with
migration
as
a
major
morphogenetic
factor.
So
then
there
are
also
models
where
they
use
migration
as
a
way
to
simulate
spatial
organization
and
differentiation.
So
migration
events
are
commonplace
throughout
development
and
pivotal
for
the
morphogenesis
of
numerous
tissues.
B
The
forces
behind
cell
motility
are
predominantly
mechanical,
but
the
signals
that
trigger
and
direct
migration
can
be
mechanical,
chemical,
electrical
or
all
three
and
so
cells
and
vivo
are
experiencing
a
multitude
of
forces
from
the
environment
and
neighboring
cells,
and
expansion
of
a
cell
population
can
lead
to
passive
movement
in
the
cells.
In
response
to
a
physic
set
of
physical
interactions,
and
so
migration
can
also
be
used
in
a
way
to
facilitate
this
sort
of
morphogenetic
organization
and
again
it
depends
on
the
system
that
you're
looking
at
in
some
systems.
B
This
is
much
more
common
in
other
systems.
Proliferation
is
much
more
common,
so
you
know
if
you're
looking
at,
like
the
a
model
of
a
of
an
organ
and
you're
looking
at
like
tumorogenesis,
one
model
might
be
much
more
suitable
than
say
another
model
like
embryogenesis
or
something
else,
and
so.
B
And
then
there's
also
differentiation
as
a
necessary
morphogenetic
factor,
so
you
can
use
differentiation
as
another.
B
Standard
for
your
model,
so
there's
a
lot
in
this
paper
and
I'm
not
going
to
go
because
we're
already
over
time.
But
I
just
wanted
to
point
this
out.
They're
also
limit
the
limitations
here.
So
there
are
a
lot
of
limitations
to
the
aging
based
model
approach,
but
I
think
it's
a
very
good
approach,
if
you're
interested
in
really
how
you
know
cells
go
from
like
an
undifferentiated
mass
to
this
highly
asymmetrical.
B
You
know
organism,
then
this
might
be
a
way
to
like
kind
of
really
get
at
some
of
those
questions.
And
again
we
have
experience
with
this
in
the
group
we've
done
morphozoic,
which
is
something
that
is
tom.
Who
was
here
earlier?
He
that
was
his
baby,
so
you
know
I
don't
know.
Maybe
we
can
do
something
with
that
going
forward.
B
We
have
the
compusell
3d,
which
we've
still
yet
to
really
kind
of
capitalize
on,
and
we
have
other
models
that
we
can
use
as
well,
and
so
these
are
all
possibilities.
This
goes
beyond
what
we've
done
with
divo
learn.
I
think
the
diva
learn
is
a
very
good
step
forward
and
I
think
maybe
the
next
step
is
to
incorporate
the
data
into
some
of
these
types
of
models.
B
So,
let's
see
what
do
we
have
here?
Susan
has
a
lot
of
comments
here
all
right,
so
development
is
not
a
reaction
to
fusion
system.
It
plays
a
role,
but
is
not
the
main
driver.
My
view
is
that
what
I've
learned
from
richard
gordon,
the
interaction
between
who's
also
in
this
group,
is
not
here
today.
The
interaction
between
the
mechanical
and
chemical
aspects
of
the
cell
and
is
not
diffusion,
so
yeah
development
is
a
reaction
of
the
living
cell
to
mechanical
and
chemical
responses
of
the
cell.
B
Many
issues
cells
cannot
think,
and
then
there
are
physical
links
between
the
force
of
the
cell
that
the
force
of
the
cell
experiences
in
nucleus.
There
is
a
physical
and
chemical
cascade
that
changes
the
dna
in
the
nucleus,
depending
on
the
forces
of
the
cell
experiences.
B
Richard
gordon
needs
to
be
here
and
add
to
what
I'm
saying
yeah
I
mean
we
can
talk
about
this
in
future
meetings,
and
so
then,
yes,
the
migration
explanation
is
more
like
what
happens
so
yeah.
I
think
that's,
those
are
good.
Those
are
some
pretty
good
observations.
B
Thank
you,
susan
again,
like
we
can
talk
more
about
this
paper
in
future
meetings.
It's
a
pretty
long
one.
So,
but
that's
that's,
like
you
know,
that's
a
way
forward.
I
think
we
can
see.
You
know
maybe
that
that'll
feel
something
interesting
and
again,
if
we're
talking
about
one
last
thing
before
we
go,
we
let's
try
to
think
about,
like
maybe
like
a
contribution
board
in
diva
learn
itself.
B
F
F
B
Yeah,
I
mean
that's,
I
think
if,
if
there's
something
well
we'll
try
to
think
about
ways
forward.
This
is
why
I
presented
the
agent-based
modeling
stuff.
If
you
want
to
contribute
something
very
specific,
we
can
talk
about
it,
but
we
should
also
probably
have
a
board.
B
I
don't
know
if
we
should
have
a
board
separate
and
diva
learn
or
if
we
should
have
a
diva
worm
or
what
I
think.
If
you're
talking
about
people
who
are
casually
engaging
with
diva
learn,
then
we
should
have
a
board
there,
but
that's
and
then
you
know
we
can
like
put
things
in
there,
and
people
can
add
things
into
that
board.
B
If
you
want
to
be
a
member
of
diva
learn,
you
can
join,
I
can
make
you
a
member,
and
then
we
can
you
know,
then
that
way,
you
can
add
to
the
board
and
do
things
like
that.
I
I
don't
know
exactly
what
to
say
about
like
contributions,
specific
contributions
right
now.
I
think
what
we
need
is
maybe,
like
some
publicity
for
this,
maybe
some
thinking
about
how
we
can
go
from
like
what
we
have
in
diva
learn,
which
is
largely
data,
analysis
and
algorithms
that
analyze
data
to
like
some.
How
do
we?
B
I
think
that's
I
mean
that's
that's
one
way
because
it's
like
you
know
we
can
propose
something,
really
grandiose,
but
it's
kind
of
like
you
know.
How
do
we
get
there,
and
so
I
think,
or
something
very,
very
there's
some.
I
think
my
oak
had
a
beginner
issue
over
the
weekend
where
there
was
something
that
was
code
related
that
could
be
fixed,
and
so
those
are
sorts
of
things
that
we
might
do.
E
Like
right
now-
or
there
is
an
issue
which
is
like
going
on
like
we
have
some
problem
in
important
connections
or
dvd
library,
which
I
believe
like
one
of
the
trick
to
do
things
right
is
just
to
try
and
accept
like
there
is
some
version
problem
in
that
library.
So
you
can
just
try
and
like
try
to
import
the
library.
If
it's
not
importing
and
showing
error,
you
can
go
to
accept
and
you
can
import
the
newer
version
of
it.
B
Okay,
yeah,
it's
yeah,
so
if
you
have
issues
you
want
to
see,
you
can
either
put
them
in
slack
or
put
them
on
make
a
board
and
we'll
probably
just
make
a
board
and
evil
learn
and
put
them
there.
B
But
we'll
we'll
try
to
organize
it
sometime
soon,
next
couple
weeks
to
get
everything
like
going
and
then
you
can
advertise
different
issues
too.
You
can
say
this.
This
issue
really
needs
to
be
addressed.
You
know
I
have
done
that
before
in
the
meeting
emails,
which
makes
it
kind
of
convenient
for
people
to
kind
of
focus
in
on
one
thing,.
E
Oh,
like
you
know
like
this
year,
like
me,
and
mike
and
jesse,
even
like
we
can
get
in
touch
with
like
like
in
that
two
of
us.
The
main
contributors
are
like
the
new
students
who
are
like
in
their
maybe
first
year
or
second
year
so
like
they
are
focused
college
groups
like
clubs,
which
are
focused
to
make
on
board
them
on
open
source,
like
in
my
college
itself.
There
is
a
club
known
as
spirit
whose.
E
B
F
B
Yeah
we
so
we're.
We
have
this
problem
that
probably
isn't
something
that
you
can
do
with
machine
learning,
which
is
to
take
a
series
of
images
that
have
been
acquired
from
something
that's
rotating
and
put
them
into
a
spherical
frame
of
reference,
and
we
talked
about
that
algorithm
a
couple
weeks
ago,
but
we're
still
kind
of
waiting
thinking
about
it,
and
so
that
might
be
something
that
we
might
go
in.
You
know
if
yeah,
usually
you
want
to
look
at
the
data
I
can.
Actually.
B
I
can
share
with
you
what
I
have.
I
have
some
of
the
data
and
I
have
a
folder
where
I
have
the
data
and
the
algorithm.
So
I
can.
I
can
send
it
to
usual
and
you
can
take
a
look
at
it
if
you
want
yeah
yeah
I'll,
send
it
to
my
oak
as
well.
So
this
is
something
yeah.
This
would
be
nice
a
little
bit
different
than
the
other
types
of
images
we're
working
with,
but
it's
definitely
something:
okay
yeah.
So
that's
good,
and
so
that's
good.
B
Let's
call
it
for
this
week
and
we'll
talk
more
on
slack
and
see
you
guys
and
have
a
good
week.
F
F
Yeah
ahead,
take
care
all.