►
From YouTube: DevoWorm (2021, Meeting 7): Positive Feedback of Bias, ResNet Voice Recognition, Multicellular Shape
Description
DevoLearn 0.3.0 release, Positive Feedback of Bias in AI, Voice Recognition using ResNet, the intellectual history of C. elegans neurodevelopment, the development of multicellular shape in soap bubbles, embryos, and cell colonies. Attendees: Susan Crawford-Young, R Tharun Gowda, Bradly Alicea, Krishna Katyal, Mainak Deb, Jesse Parent, and Shruti Raj Vansh Singh
A
B
B
B
B
B
D
B
I
think
so
yeah
just
for
today
yeah
yeah,
because
otherwise
I
I
use
a
bluetooth
microphone
to
communicate.
A
Okay,
well,
I
guess
I
don't
know
who
else
is
gonna
show
up
so
welcome
to
the
meeting
we
have
a
couple
of
things
to
talk
about
before
we
start
darren
is
going
to
give
a
demo
or
he's
going
to
show
us
some
results.
He
got
from
the
basil
area
work.
A
Okay,
that
sounds
good,
so
yeah
through
and
we'll
talk
about
the
basilaria
stuff
and
he's
been
in
contact
with
us
wall
about
it.
So
usual
is
actually
going
to
be
making
the
meetings
soon,
but
he's
got
another
conflict,
so
he
told
me
about
that.
So
he's
still
he's
still
going
at
it
he's
he's
working
with
thrown
on
this
and
then
surety
will
present
on
her
topic
and
then
I
don't
know
if
krishna
is
going
to
show
up
because
he's
we
have
some
things.
A
Maybe
to
talk
about
about
some
submissions,
but
we'll
see
so
a
couple
things
before
I
oh
go
ahead.
Oh.
B
Okay,
sorry,
thank
you
for
telling
me
about
the
gastrulation
talks.
Oh
yeah
yeah,
the
yeah,
the
one
about
the
ants
is
perfect.
I
might
be
able
to
to
use
it
somewhere
somehow
yeah
with
my
thesis
or
my
one
of
my
projects,.
A
Last
week
I
presented
on
ants
and
plants,
and
then
this
cut
last
week
later
in
the
week
there
was
a
talk,
a
gastrulation
seminar
series
that's
online
and
they
talked
about
different
types
of
embryos,
ants
and
things
like
that,
and
that's
why
I
kind
of
did
the
ants
implants,
because
I
had
seen
that
coming
up
and
I
thought
well
that'd
be
a
good
topic
to
get
into
so
susan
saw
the
gastrulation
seminar
on
that
so
that
that's
good,
it's
not
something
that
we
usually
hear
about.
E
B
Yeah,
it
was
a
different
model
organism,
one
that
they
hadn't
used.
They
said
since
the
1950s
it
was
used
prior
to
that.
So
they
thought
why
not
it's
given
them.
Some
interesting
results,
because
the
the
shell
of
the
egg
is
used
in
gastrulation.
A
Okay,
that
sounds
good
and
then
you
know
we
might
actually
talk
a
little
bit
about
that
like
following
up
on
that.
I
don't
know
if
they're
like
we
might
yeah,
we
might
go
down
a
sort
of
a
rabbit
hole
on
that.
A
If
we
have,
if
we
can
think
about
some
things,
we
might
want
to
know
more
in
depth
about
it.
I'm
not
really
familiar
with
the
model
organisms
there,
but
we
could
look
into
it
and
see
what
they
have.
A
A
So
that's
good,
so
I
guess
why
don't
I
start
off
with
a
couple
things
here.
Let's
see,
let
me
share
my
screen.
A
A
The
people
were
working
on
the
diva
learn
release,
so
this
is,
I
think,
my
knock
is
also
involved
in
this.
They
did
this
stuff
that
my
doc
was
showing
on
the
on
the
training
data,
so
they
were
able
to
expand
the
resolution
of
the
images
and
add
additional
training
data,
and
so
my
oak
combined
this
into
a
release
of
divalern.
A
Again
I
tell
them
the
rules
for
this
here
that
you
know
we
every
time
we
get
some
sort
of
major
addition.
You
know
something
that
isn't
like
a
very
small
commit
to
release
a
new
version,
and
so
he's
got
this
release.
This
was
about
seven
hours
ago,
so
it
was
pr
it's
pretty
recent
and
this
is
the
pipey
page.
So
this
is
a
pipey
where
you
can
download
open
source
platforms
written
in
python,
and
so
this
is
the
I
mean
we
still
have
it
on
github.
A
It's
just
that
this
is
the
sort
of
the
it's
like
a
a
public
repository
for
python
applications,
and
so
you
can
see
everything
is
pretty
much.
The
same
except
it's
improved.
A
With
those
changes-
and
so
this
is
congrat-
congratulations
to
everyone
involved
in
that.
A
My
oak
also
contacted
me
and
asked
me
about
different
types
of
he's
interested
in
sort
of
working
more
on
diva
learn
in
terms
of
the
c
elegans
literature.
So
I
pointed
them
into
some
papers
here
and
I'm
starting
to
create
a
bibliography
on
this.
So
people
are
interested
in
contributing
to
this
bibliography.
A
A
A
So
this
is,
you
know
three
dimensions
of
space
plus
time
and
we've
talked
about
4d
analysis
in
this
group
as
well.
So
this
is
has
a
lot
of
to
do
with
the
cell
tracking,
which
we've
talked
about
in
the
four
dimensional
analysis,
which
are
the
three
dimensions
the
space
plus
time.
So
you
have
all
these
images
where
there's
a
process
unfolding
and
that's
that's
one
folder,
and
then
we
have
biology
and
computational
biology,
which
is
this.
A
E
A
Pretty
early
it's
pretty
early
days,
but
myoke
said
that
he
found
it
to
be
pretty
useful.
So
here's
the
repository
hello,
krishna.
A
A
All
right,
so
is
this
going
over
some
of
the
news
from
this
last
week.
Yeah,
I
think
that's
it
for
now.
Well,
actually,
I
did
want
to
go
over
the
paper
submissions,
so
we've
been
talking
about
paper
submissions
and
I
want
to
go
over
this
really
quickly,
just
to
give
us
an
idea
of
like,
what's
due
soon
and
what's
not
due
soon
so
the
evolution
conference,
our
first
two
entries
on
this
list-
and
I
found
out
more
about
this
abstract
submissions
for
evolution
and
apparently
they
don't
open
until
march
1st.
A
So
I
I
guess
I
gave
the
impression
last
time
maybe
or
I
was
under
the
impression
that
the
deadline
was
march
1st.
It's
actually
the
submissions
that
open
on
march
1st.
So
I
don't
exactly
know
what
the
deadline
is,
but
that
means
that
way
more
time.
So
that's
good
news.
That
means
we
could
like
that's
when
we
could
start
submitting
things.
So
we
have
more
time
on
that.
We
have
a
couple
more
weeks.
A
Probably
we
have
this
kill
the
winners
that
that's
something
that
that
krishna
is
interested
in
and
I'm
on
that
as
well,
and
if
you
want
to
be
on
that,
we
can,
you
know,
incorporate
people's
contributions
into
that.
We
also
have
the
euler
paths
for
life.
That's
the
talk
that
I
I
showed
you
I'll.
Maybe
talk
about
the
abstract
later
on.
A
That's
also
something
for
that
conference
and
if
you're
interested
in
submitting
something
to
evolution
2021,
you
know
you
can
add
it
to
the
list
or
tell
me,
and
I
can
add
it
to
the
list,
the
flash
talk.
We
did
that
already
the
diva
learned
paper
that's
in
progress
and
we
have
those
we
have
this,
the
library
I
just
showed
you
might
play
into
that,
making
that
more
a
more
substantial
paper,
but
we'll
we'll
get
into
that.
That
has
no
deadline.
A
A
We
have
the
google
summer
of
code
stuff,
which
I
know
people
are
working
on,
and
so
those
projects
are
still,
I
don't
think
the
application
period
is
opened
yet
for
that
I
need
to
go
over
a
couple.
People
have
already
sent
me
proposals
for
that
to
review
so
I'll
review
that
I
think
my
knox
sent
me
one,
for
example,
so
I'll
review
that
and
look
it
over
and
then
I
know
we
still.
We
talked
about
having
a
separate
session
for
onboarding
for
this,
so
the
the
deadline
is
a
ways
off
on
the
application
period.
A
So
I
think
once
the
application
period
actually
starts,
which
I
think
is
at
the
end
of
this
month
at
the
beginning
of
next
month,
then
we'll
start
to
maybe
have
those
onboarding
meetings.
I
don't
know
if
we
need
to
send
around
or
we
need
to
kind
of
settle
on
a
time
date
and
time
for
that.
So
I
think
krishna
was
organizing
that
so
I
don't
know
if
he
has
an
update
on
that
or
what
his
thoughts
are.
H
D
G
G
F
F
Guess
if
a
person
is
applying
for,
he
must
be
knowing,
if
you
think
about
open
sources,.
G
And
you
know
one:
how
one
can
you.
G
H
G
G
G
You
have
to
see
the
guidelines
and
how
the
thing
was
made.
Then
you
have
to
find
a
point
where
you
can.
You
know
you
can
create
a
patch
and
you
can,
you
know
improve,
enhance
the
thing
I
can
reduce
the
bugs
and
you
can.
Yes,
you
can
fill
an
issue
and
then
submit
the
change.
G
A
Well,
I
gotta,
I
got
the
doc
like,
I
got
the
doc
attachment,
but
I
don't
have
it
open
here,
but
the
other
one.
I
don't
know
if
this
is
the
same
one
you
sent
me.
H
A
G
A
A
Write
a
paper
and
you
have
like
an
abstract,
you
know,
you
know
you
condense
it
down
one
abstract
and
that's
your
abstract
for
the
paper
and
then
you
can
submit
the
paper
somewhere
else
later.
So
I
mean
it
works
out.
Yeah.
H
J
G
Extended
abstract
for.
A
Life,
that's
the
one
that's
coming
up,
so
we
can
submit
it
there
because
we're
already
submitting
other
things
to
that.
So
that
might
be
a
good
one
and
that's
two
pages
of
their
format.
So
this
format
is
one
column.
Their
format
is
two
columns,
and
so
it's
it's
a
little
bit.
It
gives
you
a
little
bit
more
space,
but
we'll
figure
that
out
just
try
to
get
it
written
up
in
about
maybe
say
three
of
these
pages
and
then
we'll
fit
it
into
the
template.
A
Well,
yeah
we
have
to
we
have
to
kind
of
the
format
is
a
little
weird.
So
it's
not.
A
G
A
H
G
So
I
I.
G
D
G
H
G
F
Complications
or
they
will
not
respond
to
their
medication.
Even
if
there's
a
slightly,
you
can
say.
H
G
A
A
We
were
talking
about
this
there's
a
project
that
jesse
who
attends
this
group
is,
is
doing
on
ai
ethics
and
and
that
sort
of
thing,
so
we
have
an
interest
group
over
there
if
you're
interested
in
joining
in,
but
we
also
should
probably
have
a
component
of
this
in
the
diva
worm,
the
divorm
ml
materials
that
we
have
so
we
could
add
in
a
module
on
this
as
well.
I
mean
I'd
like
to
make
it
more.
A
Like
you
know,
kind
of
general,
like
you
know,
data,
you
know,
how
do
you
if
you
find
a
pattern
in
your
data?
You
know
like
for
the
the
ai
ethics
group
in
the
other
meeting.
You
know
this
is
you
know
this
is
great
for
the
divormal
I'd
like
to
focus
on
like
if
you
find
patterns
in
your
data,
you
know
how
do
you
you
know
overcome
them?
How
do
you
identify
them
if
they're
spurious
or
if
they're
so
I
mean
this-
would
be
good?
This
would
be
a
good.
F
C
C
C
G
G
A
This
brings
up
another
another
topic
too,
that
we
haven't
talked
about
in
either
either.
Group
actually
is
the
idea
of
having
like
these
human
hybrid
systems.
So,
like
you
know,
to
be
able
to
put
humans
into
the
loop
somewhere
and
to
say
you.
A
Is
this
a
good
judgment
that
the
algorithm
is
making,
or
is
it
like
something
that
I
have
a
talk
on
this?
I
haven't
presented
it
in
this
group
or
in
the
other
group,
but
I
might
present
it
at
some
point.
So,
if
you're
familiar
with
what
they
call
hybrid
chess
or
like
the,
what
do
they
call
them
the
the
man
and
and
horse
combinations,
I
can't
remember
what
they
call
them.
G
A
Yeah
yeah
yeah
so
like
in
the
90s
deep
blue
was
able
to
beat
a
human.
It
was
able
to
beat
gary
kasparov
a
chess
and
he
got
upset
about
it,
so
he
went
off
and
he
figured
out
a
way
to
like
incorporate
like
in
machine
learning.
It
wasn't
called
machine
learning
at
the
time
but
to
incorporate
that
into
playing
chess
so
like
the
human
would
supply
some
of
the
innovation
of
technique
and
the
machine
would
then
execute
the
moves.
A
And
then
that
was
like
a
different
type
of
chess
that
they
invented
in
the
aftermath
of
a
machine
beating
a
human
and,
of
course
the
human
ai
hybrids
can
do
quite
well
and
sometimes
be
machines.
A
I
don't
really
know
what
the
I
don't
really
follow
the
field,
but
like
it's,
it's
actually
an
interesting
question
and
of
course
you
can
use
this
kind
of
hybrid
systems
for
other
things
as
well
and
that's,
and
there
are
a
lot
of
advantages
to
it,
because
you
know
you
have
things
that,
obviously
humans
can
say
are
obviously
right
or
wrong,
and
machines
can't
do
that
because
they
don't
have
a
lot
of
the.
You
know
context
that
we
do
and
things
like
that.
So
it's
definitely
an
interesting
area,
topical
area.
A
E
A
So,
thank
you
for
that
krishna.
We
have
okay.
My
knock
said
nice
presentation
krishna.
So
let
me
share
my
screen
again.
So
we're
going
back
to
the
paper
submissions.
So
that's
that's
krishna's
stuff,
so
he's
got
the
evolution.
Abstract
he's
got
some
other
things.
I
don't
know
if
I
put
the
a
wife
submission
on
this.
A
I
have
another
document
from
my
other
group
and
I
may
have
put
it
on
there,
but
let's
just
pencil
in
this
abstract
that
krishna
showed
the
two-page
abstract
on
the
bnns
and
a
nns
yeah.
I
won't
put
it
on
this
list
to
have
it
on
the
other
list,
though
so
that'll
be
submitted
to
a
life
conference
and
in
the
tentative
deadline
for
that
is
march
7..
A
So
that's
coming
up,
but
you
know
we
can
it's
not
that
much.
I
think
you've
got
a
lot
of
it
in
there.
We
just
need
to
revise
it
a
bit
and
maybe
next
week
we'll
we'll
go
over
it
a
little
bit
more
detail
and
kind
of
walk
through
it
a
bit
more
so
then
we
have
this.
Then
we
have
some
papers
that
we
want
to
submit.
A
So
there's
this
bacillary
non-renal
cognition
paper,
that's
something
that
has
been
accepted
as
a
book
chapter
in
the
proposal
form
and
then
it's
due
april
30th
and
I
think
we're
far
far
along
on
that.
But
we
we
are
going
to
turn
our
attention
to
that
in
the
coming
weeks
and
you
know
kind
of
fill
up
fill
in
the
gaps,
because
I
think
we
have
a
good
like.
A
I
think
we
have
a
good
basis
for
everything
like
everything's
in
place
for
it.
It
just
needs
to
be
like
fleshed
out
into
a
actual
chapter.
The
c
elegans
conference
is
coming
up
on
the
twi
or
the
deadline
for
that
is
coming
up
on
march.
25Th.
Excuse
me,
and
this
is
the
submission
portal.
It
will
let
you
know
that,
like
they're,
they
charge
a
fee
for
submitting
an
abstract
and
there's
a
fee
for
attending
the
conference,
but
it's
totally
online.
A
A
The
the
euler
paths
for
life-
that's!
This
is
the
abstract
that
I
want
to
submit
to
evolution
and
that's
more
of
a
biological
abstract
and
then
there's
this
full
paper,
which
is
to
be
submitted
to
complement,
which
is
a
network
science
conference
and
that's
the
26th
of
march.
A
A
Okay,
so
the
way
I'm
doing
this
for
this
paper
is,
is
euler
cycles
for
life,
and
I
might
kind
of
go
over
this
really
quickly
later,
like
the
slides
to
remind
you
what
it's
about,
but
I
have
two
versions
of
this:
one
is
an
evolution
of
development
approach,
which
is
more
like
biological
language
and
a
more
biological
motivation,
and
then
the
network
science
version,
which
is
more
network
science
like
so.
If
we
go
into
the
evolution
of
development
version,
we
see
that
there's
a
you
know.
A
So
there's
a
focus
on
like
the
functional
aspects
of
these
these
networks
and
then
tracing
the
networks
with
these
euler
cycles
and
again
I'll
talk
about
what
I'll
kind
of
briefly
go
over
what
that
is
later
on,
but
we're
going
to
use
euler
pads,
which
are
a
tool
from
graph
theory
to
evaluate
these
networks
that
exist
just
between
the
edges
of
cells.
So
you
know
you've
seen
the
embryos,
for
example,
they
have
cells
and
the
cells
form
junctions
between
each
other.
Well,
they
form
boundaries
between
each
other
and
along
those
boundaries.
A
There
are
all
sorts
of
gap,
junctions
and
other
passageways
between
cells
and
then
we're
evaluating
those
edges
to
see.
If
you
can,
you
know,
form
a
continuous
network
which
implies
that
there's
some
sort
of
like
equilibrium
network
for
transferring
things
from
the
edges
of
the
multi-cellular
aggregate
into
the
middle
of
the
aggregate.
So
that's
that's
basically
the
motivation
here
and
then
this
kind
of
goes
into.
A
You
know
how
you
can
sort
of
simulate
by
doing
this
exercise
simulate
the
establishment
of
new
phenotypic
modules,
so
there's
an
emphasis
on
modularity,
but
then
the
network
version,
which
is
a
full
paper,
focuses
a
little
bit
differently
and
that
is
again
we're
trying
to
form
these
networks
motivated
a
little
bit
with
biology.
A
But
then
we
kind
of
get
into
this
kind
of
use.
Some
graph
theory
language
here
to
describe
the
graphs,
and
you
know
this
kind
of
describes
the
mechanism
and
this
actually,
the
the
the
thing
we're
gonna
focus
on
in
this
abstract
is
where
you
have
these
meta
replication
events,
where
you
have
a
unipart
type
network
that
becomes
a
bipartite
network
which
is
different
than
sort
of
the
assumption
of
a
module,
phenotypic
module.
A
But
it's
it's
very
similar,
but
I'm
going
to
kind
of
use
the
I'm
going
to
use
different
types
of
language
with
these
different
submissions.
So
I'm
what
I'm
trying
to
show
you
here
is
that
when
you
submit
something
to
different
venues,
sometimes
you
have
to
change
the
language
a
bit
to
get
people.
You
know
to
get
people's
attention
to
make
it
relevant,
but
you
know
that's
that's!
So
that's
that's
over
how
that's
that's
gonna
proceed.
A
The
app
simply
just
submitting
an
abstract
of
course
means
that
you're
going
to
create
some
slides
later,
so
the
slides
are
going
to
focus
on
the
biology,
in
this
case,
for
the
evolution
conference,
but
beyond
network
science
for
the
complement
conference
and
graph
theory
too.
So
that's
that's
my
lesson.
For
the
day.
On
that,
so
I
also
have
this
other
idea,
and
I've
submitted
this
to
networks
20
2021,
which
is
a
it's
also,
a
virtual
conference.
It's
network
theory,
it's
I
don't
know.
If
I
have,
I
don't
have
a
link
to
the
abs.
A
A
This
is
about
embryo
networks
is
generative
divergent
integration.
So
we
have
these
things
called
embryo
networks
which
you
published
on
a
couple
years
ago
and.
A
This
to
nutsai
a
couple
years
ago
as
well
to
a
conference,
and
the
idea
is
that
you
have
these-
that
embryos
from
these
networks.
If
you
look
at
the
different
cells
in
an
embryo,
they
interact
with
one
another
and
they
form
these
networks
that
you
can
define
through
graph
theory
using
thresholds
and
they
form
these
complex
networks,
and
the
idea
is
that
they
form
networks
and
they
form
structure
over
time
as
cells
divide
and
move
around
in
the
embryo,
and
they
form
these
modules,
as
I
mentioned
before,
they
form
phenotypic
modules.
A
They
form
like
highly
densely
connected
cells,
meaning
that
they're
cells
that
are
very
tightly
packed
together.
And
so
all
these
things
happen,
but
in
say
like
something
like
c
elegans.
You
have
these
developmental
cells
doing
this,
and
then
you
also
have
a
nervous
system,
that's
starting
to
form.
A
So
the
nervous
system,
of
course
forms
this
connectome,
which
we
all
know
from
the
c
elegans
connectome
literature
that
you
have
like
the
basically,
as
the
neurons
begin
to
differentiate
and
form,
you
know,
functional
neurons,
they
form
these
neural
networks
that
have
connections
between
them,
and
so
what
I'm
talking
about?
In
this,
abstract
is
the
formation
of
these
two
networks
from
a
single
network
so
early
on
in
in
embryogenesis.
A
You
have
this
one
network,
that's
basically
the
proximity
of
different
developmental
cells
and
maybe
some
gap
junctions
in
between
them
that
form
connections,
because
they're
neighbors,
but
also
these
you
get
start
to
get
like
sets
of
neural
cells
that
are
going
to
later
function
together
into
a
neural
network
into
a
connectome.
And
so
you
get
this
divergence
into
two
different
networks,
and
so
this
this
abstract
talks
about
this
and
this
process
and
highlighting
this
process.
So
this
is
going
to
be
a
talk,
maybe
at
netsigh
and
probably
also
an
abstracted
complement.
A
So
that's
that's
something
we
can
talk
about
later.
I
just
wrote
that
abstract
I
think
this
week,
so
this
is
there's
more
to
come
on
that
more
more
immediate,
that
we
have
the
biosystem
special
issue,
that's
on
march
1st,
so
we
have
the
we
have
a
couple
papers
for
that.
We
have
the
the
temporal
paper
the
bursts
paper,
where
we're
talking
the
one
we've
been
talking
about
with
c
elegans
and
zebrafish,
and
that
is
there
are
some
trivial
revisions
to
that.
A
So
I'll
probably
go
through
that
and
again,
if
you
want
to
contribute
to
this,
one
way
you
can
is
by
finding
a
good,
open
access
or
fair
use,
animation
of
of
of
embryogenesis,
especially
like
a
time
series
sort
of
thing.
So
that's
what
I'm
trying
to
find
now.
B
See
it
takes
three
images
from
three
dimensions
and
compare
how
this
compares
with
previous
attempts.
A
B
Including
my
flipping
microscope,
so
I
still
don't
have
the
the
3d
software
to
project
all
of
this
onto
a
sphere.
B
A
A
I'll
just
put
that
there
instead,
okay,
so
I'll,
put
that
yeah
I'll
put
this,
so
anyone
contributing
anything
to
that
the
deadline
is
march
1st,
so
it's
coming
up.
I
know
that
oswald
and
jesse
are
currently
co-authors
on
it.
So
I
wanted
them
to
get
a
heads
up
on
it.
We'll
we'll
talk
about
it
more
this
week,
then
there's
this
amazon
book,
which
we
talked
about
last
week.
A
That
was
something
that
krishna
was
spearheading
and
we'll
talk
more
about
that
in
coming
weeks
and
then
finally,
there's
this
incf
neuroinformatics
assembly
coming
up-
and
this
is
the
incf.
These
are
our
sponsors
for
google
summer
of
code,
but
they
also
have
this
assembly
coming
up
and
we
need
a
1
500
character
abstract
for
that.
A
So
I
think
one
of
the
things
you
might
focus
on,
for
that
is
maybe
some
sort
of
let's
see,
yeah
some
sort
of
diva
or
devo
zoo
epistemological
directories
mash
up,
and
if
you
don't
know
what
epistemological
directories
are
it's
a
thing
that
we're
working
on
in?
In
my
other
group,
I
I
think
if
we
do
that
I'll
I'll
kind
of
maybe
give
a
presentation
about
that
in
this
group
a
little
bit
to
give
you
an
idea
of
what
this
is
and
then
you
know,
write
up
an
abstract
so.
J
A
You're
interested
in
in,
but
I
this
isn't
for
sure
I
don't
know
what
we're
gonna
do
for
this,
but
if
you
want
to
participate
in
this,
let
me
know
I
think
this
is
a.
This
will
be
a
good
opportunity
for
incf
to
see
what
we're
doing.
I
know
we
I
presented
last
year
at
the
gsoc
mentors
a
workshop,
and
so
they
were
pretty
excited
about
that.
So
that's
a
good
opportunity
as
well.
A
So
let
me
unshare
my
screen
now
and
let
me
look
at
the
chat.
Okay,
yeah
krishna
says
I
had
jokes
to
catch
attention.
Those
are
your
memes
that
you
put
in
your
talks,
they're
very
nice,
so
shruti
do
you
want
to
now
present
on
your
topic.
F
I
Model,
religion,
something
which
had,
which
is
worked
on
using
machine
learning,
with
deep
learning.
I
C
C
I
A
C
Maybe
so
yeah
the
data
contains.
Maybe
the
next
three
audios
are
a
little
more
louder,
so
yeah
the
the
data.
This.
Obviously
the
kids
are
talking
about
the
door
and
what.
I
C
C
C
C
Okay,
so
yeah.
This
was
like
the
audio.
F
C
C
You
know
plotting
the
frequency
in
on
the
timelapses
would
help
the
model
enable
the
model
to
recognize
the
emotions,
and
I
used
resonant
50
for
training
my
models.
I
fast,
I
first
of
all
converted
my
audio
into
images
like
I
showed
you
before,
and
then
those
were
trained
in
the
model
and
I
got.
A
F
E
F
F
Script,
that
will
run
with
like
it
will
capture
the
voice.
F
C
B
Thank
you
for
the
talk
that
it
was
interesting.
I
think
that's
a
very
interesting
line
of
research
and
I
I
have
noticed
when
raising
children,
that
when
things
got
quiet,
that's
when
you
had
to
worry
about
things.
Yes,
that
is
what
it
does.
C
A
A
And
then
people
can
look
at
it
and
go
back.
It's
very
good
work.
Thank
you.
So
I
I
know
jesse
missed
out
on
krishna's
talk
on
the
ai
ethics
or
the
positive
feedback
thing.
So
I
don't
know.
Maybe
krishna.
Can
you
put
your
slides
in
the
slack
as
well
to
see
technically.
C
I
guess
he
would
have
to
see
the
whole
meeting
today.
Oh.
A
C
A
At
that
too,
but
so
yeah
yeah,
you
missed
it
jessie,
I'm
gonna
go
back
and
I'm
gonna
do
some
things
if
you
have
to
leave
at
the
top
of
the
hour
go
ahead,
but
I'm
gonna
go
through
some
things
here,
I'll
go
through
our
papers,
but
I
also
wanted
to
talk
a
little
bit
about
this.
A
very
short
thing
on
this.
A
Euler
cycles
for
life
here,
so
let
me
go
into
this,
so
these
are
I
I
presented
this
a
long
time
ago
and
I
don't
think
maybe
people
remember,
maybe
they
don't
so.
This
was
a
talk
from
I
think.
Last
year,
the
year
before
that
I
gave
in
the
meeting
and
I'm
not
going
to
go
through
the
whole
talk.
A
I
just
wanted
to
point
out
as
a
refresher
what's
going
on
here,
so
the
idea
is
that
euler
paths
are
similar
to
the
the
kenningsburg
bridge
problem,
so
the
cannings
work
bridge
problem
was
a
problem
that
was
posed
in
about
the
18th
century.
In
germany,
where
there
was
this
town
kenningsburg
and
it
had
like
seven
bridges
that
you
know,
because
there
was
like
an
island
and
another
island.
F
A
Here
and
some
mainland
and
they
had
a
bunch
of
bridges
and
the
idea
was
how
do
you
cross
each
bridge
only
once
you
know
and
still
make
the
entire
circuit,
like
you
know,
and
and
this
kind
of
is
similar
to
the
traveling
salesman
problem,
where
you
have
to
visit
a
bunch
of
cities
in
the
most
efficient
way.
A
So
that
means
only
like
going
down
one
path
once
instead
of
doubling
back
and
so
that's
basically
the
idea
here,
an
euler
cycle
is
a
way
to
analyze
these
paths,
to
make
sure
that
you
find
the
most
efficient
path,
and
so
I
put
these
seashells
and
diatoms
in
here
because
as
you'll
see,
this
is
like
the
kind
of
phenotype
that
we're
kind
of
interested
in,
but
also
things
like
volvox,
which
is
a
single
cell
colony,
remote.
It's
a
multi-cell
county
of
single
cell
organism.
A
So
it's
this
it's
another
phenotype
that
might
be
relevant
to
this
approach
and
again
you
see
euler
paths
in
terms
of
like
geometric
shape.
So
this
is
an
octahedron.
This
is
a
dodecahedron,
so
you
have
a
lot
of
like
potential
see
these.
These
partitions
here,
like
kind
of
like
cells
and
then
these
blue
lines
are
the
paths
between
the
cells
and
so
on.
On
dodecahedron,
you
have
a
number
of
cells
that
maybe
form
a
complex
shape
that
has
some
depth
to
it,
but
you
have
these
these
blue
lines
and
those
are
the
networks.
B
Art:
oh,
go
ahead!
Sorry,
are
you
going
to
look
at.
B
B
A
I
don't
have
those
in
here,
but
that's
a
good
point.
I
didn't
bring
that
in
and
there
are
some
like
optimization
problems
having
to
do
with
soap
bubbles
which,
if
you
ever
looked
at
like
like,
if
you
ever
suds
up
a
a
sink
of
water
or
a
bathtub,
you
see
like
the
suds
form
and
then
the
these
bubbles
that
form
clusters
and
they
form
these
sort
of
hard
edges
inside
that
are
not
round
they're
kind
of
square
like
and
that's
a
kind
of
a
soap
bubble.
A
A
cluster
and
those
actually
are
very
similar
to
this
thing
as
well-
and
I
don't
have
the
slides
in
here-
but
that's
a
very
there's-
some
optimization
problems
surrounding
soap
bubbles
that
if
you
look
in
the
literature,
you'll
find
and
so
that's
relevant
as
well.
But
why
are
cell
boundaries
important
in
living
systems?
And
so
we
have?
These?
All
sorts
of
things
happen
at
the
interface
between
cells.
We
have
these
gap
junctions
that
allow
for
electrical
communication
between
cells.
A
So
we
know
you
know
you
might
be
familiar
with
synaptic
connections
between
two
cells,
but
those
are
those
are
chemical,
but
these
are
electrical
connections
and
they
connect
when
two
cells
are
near
each
other
and
there's
this
connection
that
crosses
the
bound
allows
things
to
cross
the
boundary
between
the
two
cells.
A
It's
a
it's
a
different
form
and
it's
used
in
the
nervous
system.
It's
used
in
all
different
types
of
cells.
So
that's
one
reason
why
we're
interested
in
these
boundaries.
Another
is,
for
you
know,
different
types
of
nutrients,
sort
of
a
protovascular
system
that
goes
in
between
the
cells
in
plants.
You
have
this
movement
of
transcription
factors
and
micrornas
that
happens
between
plant
cells
and
so
that's
something
that's
important
in
plants.
A
A
You
can
only
cross
each
edge
once
and
that's
the
idea,
and
if
you
cross
it
twice,
then
it's
no
longer
a
unified
network,
it
actually
bifurcates
into
two
networks,
and
you
have
to
split
it
into
two
networks
and
analyze
them.
So
the
idea
here
is,
we
want
to
find
an
euler
circuit
and
that's
how
we
define
a
unified
system
and
if
we
have,
if
we
can't
do
that
in
some
cases
you
can't
solve
certain
networks.
A
You
can't
find
a
complete
euler
circuit,
then
you
have
to
divide
it
up
into
two
networks,
and
so
we
can
actually
use
a
tool
called
graph
online
to
calculate
these.
So
we
give
it
a
network
and
then
it
you
know
we
define
the
nodes,
we
define
the
edges
and
then
it
evaluates
the
edges
pretty
quickly
and
it
tells
you
whether
there's
nuclearian
path
or
not,
and
so
the
idea
would
be
that
in
evolution
you're
adding
these
shapes,
so
you
have
triangles
here.
A
You
start
with
one
triangle,
you
add
another
triangle,
you
add
another
triangle
and
eventually
you
know
you
you'll
keep
getting
a
complete
euler
path
or
in
some
cases,
when
you
add
in
one
another
shape
and
you
keep
replicating
it.
You
end
up
with
an
incomplete
euler
path
and
in
that
case,
then
it
divides
into
two
different
modules.
And
so
then
you
start.
You
start
to
have
two
different
sub
modules
that
have
to
exist
separately.
A
That's
the
way,
we're
that
those
are
the
rules
of
defining
this,
so
in
this
case
this
network,
which
is
actually
a
collection
of
different
shapes,
they're
triangles
here,
but
they're
also
squares.
This
does
not
give
you
a
eulerian
path,
and
so,
depending
on
your
generation
rule
for
this,
you
get.
You
can
end
up
with
these
things
that
if
you
know,
if
we
go
back,
go
back
a
few
steps
and
take
away
some
of
these
edges,
we
can
have
a
eulerian
path
so
there.
This
is
how
this
works.
A
You
start
off
with
a
very
simple
shape.
You
start
adding
on.
You,
maybe
replicate
a
shape,
or
you
add
in
different
shapes,
and
you
end
up
with
something
that
has
to
be
separated
into
two
different
modules,
and
then
you
keep
growing
a
more
and
more.
You
know
complicated
phenotype,
so
you
know
you
end
up
with
something
like
a
seashell
or
a
diatom,
which
is
a
very
complicated
shape.
It
has
a
lot
of
different
elements
in
it.
Some
are
you
know
triangular.
Some
are
circular,
some
are
square,
but
they
all
have.
A
This
sort
of
you
know,
structure
that
we
can
evaluate
like
this,
and
so
we
can
also
look
at
like
things
like
rotation
and
scaling,
and
so
in
this
case
we
have
a
graph
that
forms
a
eulerian
path,
but
then
we
can
distort
it
and
actually
form
different
types
of
different
phenotypes
that
are
much
more
complex
than
the
original,
and
then
that
also
may
change
the
transformation
rules
in
terms
of
like
what
gets
replicated.
A
So
we
can
actually
do
this
in
a
way
that
replicates
more
complex
phenotypes,
and
so
this
is
just
a
simple
transformation
rule
here,
where
you're
moving
one
node
and
you're,
stretching
it
out
and
you're
stretching
the
entire
network
with
it.
And
so
the
idea
here
is
to
like
evaluate
very
simple,
geometric
shapes
and
geometric.
A
You
know,
sort
of
the
you
know,
replication
looks
for
different
cells
and
then
stretch
it
out,
like
you
might
see
in
morphogenesis
to
get
these
even
more
complex
shapes,
but
we
can
still
evaluate
it
using
this
simple
methodology,
and
so
I
that's
all
I'm
going
to
talk
about
now.
I
was
going
to
get
into
artificial
genomes
and
hypercubes,
but
that's
that's,
I
think,
a
bit
much.
A
I
can
go
back
to
that
later,
so
I
wanted
to
just
bring
that
up,
because
I
wanted
to
refresh
people's
memories
on
the
origins
of
that
work,
and
so
it's
still
a
work
in
progress.
Let's
say
I
want
to
get
out
these
submissions
because
I
want
to
like
improve
that
work
somewhat.
A
If
you
have
any
ideas
on
how
to
improve
it,
let
me
know
I've
been
kind
of
working
on
it,
like
kind
of
in
the
background
a
bit,
so
I
haven't
spent
a
huge
amount
of
time
on
it,
but-
and
I
I
I
started
working
on
a
nervous
about
two
years
ago,
so
I
think
I'm
going
to
go
over
maybe
two
or
three
papers
here.
A
The
first
one
here
is
this
paper
in
the
journal
of
neurogenetics,
and
this
is
a
perspective
on
c
elegans,
neurodevelopment
from
early
visionaries
to
a
booming
neuroscience
research,
and
this
is
a
paper.
That's,
I
think,
more
of
a
historical
point.
A
It
just
kind
of
goes
into
like
how
c
elegans
has
been
how
they
started
using
c
elegans.
As
a
neurodevelopmental
tool,
so
people
will
use
c
elegans
to
look
at
nervous
systems
that
are
developing.
Looking
at
like
the
different
things
that
are
expressed
in
different
cells.
We've
talked
about
how
c
elegans
we
know
every
cell
and
its
function
and
in
development.
That's
no
less
the
case.
A
We
have
a
lineage
tree
where
we
know
every
developmental
cell
and
then
what
its
ultimate
fate
is
going
to
be
that's
what
they
call
deterministic
evolution
or
deterministic
development,
where
we
know
if
we
follow
a
cell
lineage
what
that
cell
is
going
to
become,
whereas
in
in
mammals
you
get
cell
lineages,
where
you
get
stem
cells
that
proliferate
and
then
they
respond
to
local
signals
to
make
their
the
to
differentiate
into
functional
cells,
and
so
it's
easy
in
c
elegans,
because
not
only
do
you
have
that
deterministic
system,
but
you
also
have
the
ability
to
trace
through
and,
and
you
have
common,
you
know
things
in
common.
A
You
have
homologous
neurotransmitters
and
things
like
that,
so
you
can
actually
look
at
things
like
diseases
and
you
can
look
at
those
aging.
For
example,
you
can
look
at
all
those
things
and
see
elegance
and
you
know
get
all
the
advantages
of
the
c
elegans
model
organism,
but
also
with
the
applica
applicability
to
humans,
and
so
it's
you
know
they
talk
about
this.
In
this
paper
c
elegans
has
been
mapped
at
the
level
of
genes
cells
and
synapses,
providing
the
first
meta
zone
with
a
complete
cell,
lineage
sequence,
genome
and
connectome.
A
I've
pointed
out
recent
noteworthy
findings
in
the
fields
of
glia
biology,
sex,
dimorphism
and
plasticity,
and
neural
development,
highlighting
our
current
research
connects
back
to
pioneering
studies
by
brenner,
sulston
and
colleagues,
so
sidney
brenner
was
the
first
person
to
propose
using
c
elegans
as
a
model
organism
and
before
that
it
was
a
very
obscure
mo
an
obscure
organism.
I
mean
you
know,
people
wouldn't
really
think
much
of
nematodes.
A
John
john
solston,
of
course,
was
the
first
one
to
map
out
the
cell
lineage
tree
of
c
elegans,
and
so
he
sat
there
looking
at
a
c
elegans
under
a
microscope
and
actually
drawing
it
out
by
hand
which
seems
you
know
up
like
a
lot
of
work
and
it
was,
but
it
gave
us
a
lot
of
you
know
it
gave
us
a
good
basis
for
a
lot
of
good
science.
A
So
this
review
kind
of
goes
through
this.
It
gives
you
a
timeline
of
the
20th
and
21st
century
in
terms
of
some
milestones
in
in
c
elegans
neurodevelopmental
research.
So
if
we
zoom
in
here
to
this
timeline,
we'll
see
that
there's
in
1973
you
get
the
first
genetic
screen
and
what
they
call
defined.
A
Mutants
people
have
investigated
what
they
call
the
post-embryonic
cell
lineage,
looking
at
some
structural
biology,
looking
at
some
anatomy
and
then
in
in
the
80s
we
got
to
the
embryonic
cell
miniature,
which
was
finalized
in
the
early
80s
in
84
we
had
a
physical
genome
map
and
then
in
93
people
start
using
gfp,
which
is
that
green?
A
You
know
fluorescent
dye
or
the
green
fluorescent
signal
that
people
use
to
look
at
different
things
in
the
cell,
so
you
can
express
gfp
with
some
gene
and
look
at
how
it's
expressed
in
a
cell.
So
all
these
things
came
together
like
kind
of
gradually
and
then
all
at
once.
We
had
this
great
model
for
looking
at
different
aspects
of
the
biology
there
is.
You
know
people
were
starting
to
investigate
the
mechanisms
of
synaptogenesis
in
development,
neuron
rules
for
animal
behavior,
so
c
elegans.
A
But
you
know
c
elegans
gives
you
that
ability
and
then,
of
course,
going
into
the
21st
century,
you're
getting
whole
genome
sequences
like
of
the
whole
genome,
so
it
makes
it
easier
to
understand
the
what
the
defined
mutants
are
all
about
their
location
in
the
genome
and
everything
and
then
mapping
it
to
humans,
because
we
want
to
know
you
know
for
some
of
the
disease
research,
how
it
maps
back
to
humans
in
the
human
genome.
A
So
there's
a
lot
of
really
cool
stuff
on
this
time
timeline.
There's
the
male
connectome.
A
couple
years
ago,
glia
born
neurons
in
2014
nervous
system
maintenance
back
in
around
2010,
which
is
like
you
know.
How
do
you
keep
the
brain?
How
do
you
keep
the
nervous
system
from
being
overwhelmed
by
different
proteins
you're,
where
alzheimer's
disease
is
caused
by
an
aggregation
of
plaques
and
tangles
of
different
proteins
that
accumulate
in
the
brain
and
so
in
the
nervous
system
of
c
elegans?
A
You
have
similar
issues
with
you
know:
maintaining
the
nervous
system
and
and
the
cells
and
and
their
healthy
state.
So
those
things
are
really
relevant
to
disease.
They've
done
a
lot
of
studies
with
micro
rnas
in
terms
of
studying
neuron
fate,
so
they've
kind
of
refined,
the
you
know
how
neurons
become
sort
of
differentiated
cells.
You
know
it's
it's,
you
know
they're
deterministic,
but
there's
also
some
variation
in
their
function,
based
on
like
things
like
micrornas
and
other
types
of
neurotransmitters,
and
things
like
that.
A
So
this
paper
it
kind
of
really
goes
over
a
lot
of
this
research
and
if
you
go
to
the
c
elegans
meeting,
if
you
attend
that
you'll,
there's
like
a
very
strong
appreciation
for
history
in
that
meeting,
where
they
talk
about
some
of
these
names
that
you'll
read
about
in
this
paper.
So
they
talk
about
solstin
and
horvitz.
They
talk
about
brenner.
They
talk
about
some
marty
chalfie
and
some
of
these
other
people,
who
really
you
know,
did
a
lot
of
the
sort
of
the
groundwork
for
understanding
c
elegans.
A
Is
this
kind
of
you
know
the
the
sort
of
the
complete
understanding
that
we
have
or
somewhat
complete
understanding
we
have
of
it
today.
So
this
is
a
nice
graphic
here
that
shows
development
in
terms
of
cell
birth
and
diversification
path.
Finding
of
neuro,
you
know
neurons
when
they
send
out
their
axons.
A
They
do
pathfinding
to
find
other
cells
to
connect
to,
and
then
they
have
this
synapse
formation
diagram,
which
shows
a
lot
of
the
synapse
formation
that
happens.
So
I
actually,
I
wrote
a
paper
about
this
summarizing
this
as
well,
and
this
is
a
this
is
a
good.
You
know
we
have
it
down
to
like
specific
times
for
these
events.
A
So
a
lot
of
the
wiring
of
the
a
chemical
connectome
happens
after
the
egg
hatches,
so
the
egg.
You
know
you
get
this
this
morphogenesis
within
the
egg
and
then
it
hatches-
and
then
you
get
this
this
finalization
of
the
chemical
connective
or
the
synapse
connections
during
it
within
the
egg.
There's
actually
a
lot
of
this
electrical
connection
between
the
gap
junctions,
but
there's
also
a
lot
of
this
positioning
of
cells
and
migration
of
axons,
and
things
like
that.
F
A
Just
so,
you
have
a
timeline
for
that,
and
so
this
kind
of
goes
through
the
nervous
system,
formation
and
history.
So
this
is
a
very
good
paper
if
you're
interested
in
that,
I
would
definitely
read
it
I'll
go
through
this
paper.
A
A
This
is
a
c
elegans
relevant
paper
as
well.
So
this
is
about
something
called
blastomere
packings
and
the
abstract
reads
that
at
verily
early
embryonic
stages,
when
embryos
are
composed
of
just
a
few
cells
establishing
the
correct
packing
arrangements
or
contacts
between
cells
is
essential
for
the
proper
development
of
the
organism.
A
As
early
as
the
4
cell
stage,
the
observed
cellular
packings
in
different
species
are
distinct
and
in
many
cases,
differ
from
the
equilibrium
packings
expected
for
simple
adherent
and
deformable
particles.
So
this
is
about
the
physics
of
the
early
embryo
and
how
the
cells
are
arranged
and
then
explaining
context
between
those
cells
in
that
way.
So
we
you
know-
and
this
is
not
just
a
random
thing-
this
is
very
specific
to
the
function
of
the
cell
and
the
later
parts
of
embryogenesis.
A
Here
we
simulate
the
non-equilibrium
dynamics
of
cells
in
early
embryos
and
systematically
study
how
these
different
parameters
affect
embryonic
packings
at
the
four
cell
stage.
In
the
absence
of
embryo
confinement,
we
find
that
the
cellular
packings
are
not
robust
with
multiple
packing
configurations
simultaneously
possible
and
very
sensitive
to
parameter
changes.
A
A
Indicate
that
physical
confinement
of
the
embryo
is
essential
to
robustly
specifying
proper
cellular
arrangements
at
very
early
development
developmental
stages.
So
this
is
something
that
I
think
susan
would
be
interested
in
following
up
on.
Maybe
she
is
interested
in
a
lot
of
the
stuff
related
to
embryophysics
and
like
the
the
sort
of
the
packings
of
cells,
as
I
recall,
so
this
is
a
this
paper
kind
of
gets
you
down
to
the
very
early
stages
of
embryogenesis,
something
they
call
the
blastomere.
A
So
it's
really
when
you
have
just
these
founder
cells
of
the
different
lineages
packed
together,
and
so
you
can
study
this
in
c
elegans.
So
in
the
nematode
c
elegans
they
sort
of
form
a
diamond
shape
in
mouse
embryos.
They
form
a
tetrahedron
and
in
sea
urchin
cells,
arrange
in
a
square
configuration
and
so,
depending
on
this
organism,
you
go
to
even
at
that
stage
they
form
a
certain
geometry,
and
it's
not
necessarily
this
the
same
one
you're
going
to
say
something:
okay,.
B
B
A
A
In
the
I
mean,
I
think
that
the
I
think
the
assumption
is
is
that
it's
just
that
there
isn't
much
to
do
with
shape
that
early
on
in
embryogenesis,
but
we're
I
guess,
they're
finding
that
that's
actually
not
really
true,
that
there's
a
lot
of
variation,
but
they
did
they
do
a
lot
of
like.
I
think
they
do
a
lot
of
simulation
in
this
paper.
They
kind
of
consider
the
contact
angle
and
all
these
other
things
that
are,
and
they
look
at
mouse
c,
elegans
and
sea
urchin.
A
So
they're
they're
doing
this
wide
range
of
phylogeny
here.
So
you
can
see
that
what
they
mean
by
the
the
angles
here.
So
the
c
elegans
you
have
this
diamond.
So
you
have
two
cells
here
top
and
bottom.
You
have
one
that's
anterior
and
one
that's
posterior
and
then
in
mouse
you
get
this
triangle
shape
and
then
yeah
and
then
sea
urchin,
it's
squared
off.
So
you
actually
in
in
sea
squirts,
I
know
from
some
of
our
earlier
work.
They
divide
they
sort
of
bud
into
two
cells.
A
They,
you
know
they
have
this
sort
of
two
cell.
They
just
sort
of
replicate
and
then
stay
in
place.
They
don't
really
move
like
anterior
posterior,
like
c
elegans.
A
They
just
have
this
sort
of
square
packing
shape
like
this
it's
very
square
and
then
they
also
have
a
lot
of
asymmetric
divisions
so
like
in
the
human
case
you
mentioned,
and
then
they
form
you
know
later
on
they're,
going
to
form
this
pretty
complex
morphology,
but
in
this
stage
you're
just
doing
it
in
squares,
like
you
know,
they're
just
going
in
one
orientation
at
a
time.
So
it's
really
interesting.
If.
B
Yeah
an
interesting
stage
in
vertebrate
embryos
is
the
part
where
they
pack
together,
or
I
guess.
B
Embryos
for
memorial
and
that's
when
they
cells
packed
together,
they
kind
of
cramped
themselves
together
before
they're,
more
loose,
loosely
associated.
A
A
This
is
really
good
paper.
It's
it's
actually
an
uncorrected
proof,
so
there
might
be
the
actual
corrected
proof
might
be
out
by
now,
because
I
think
I
found
this
paper
a
couple
weeks
ago
and
put
it
in
the
list.
So
this
is
a
and
I'll
provide
a
link
to
this
later
on
this
folder.
E
A
That
one
I'll
talk
about
that.
Well,
I
think
krishna
had
to
leave
but
there's
a
paper
on
hub
connectivity,
neuronal
diversity
and
gene
expression
in
the
c
elegans
connectome,
and
that
might
be
something
interesting
for
the
difference
between
artificial
neural
networks
and
biological
neural
networks,
because
artificial
neural
networks
are
these
representations
that
just
have
these
cells
and
they're
connected,
but
they're,
just
we're
kind
of
approximating
just
the
electrical
or
the
chemical
activity
between
the
cells?
And
so
you
know
we're.
A
That's
all
the
only
thing
we're
really
interested
in
but
of
course,
nervous
systems
are
much
more
detailed
than
that.
They
have
a
lot
of
things
going
on.
They
have
these
different
motifs
in
their
connectomes,
and
one
of
those
is
this
hub
connectivity,
which
is
where
you
have
certain
neurons,
that
connect
to
many
other
neurons
and
form
functional
hubs,
and
so
you
have
this
they've
looked
at
different
brains,
not
just
c
elegans,
but
mouse
and
humans,
and
so
you
can
find
these
hubs
in
any
kind
of
nervous
system
you
want.
A
They
call
them
rich
club
neurons
because
they're
virtually
connected
relative
to
other
neurons
and
so
again
these
are
these
chemical
connections,
and
so
in
this
case
they
look
at
c
elegans
connectome.
That
look
to
look
at
like
how
gene
expression
is
coupled
to
this.
These
differences
in
connectivity,
and
so
they
looked
at
279
c
elegans,
neurons
and
binary
gene
expression,
data
for
each
neuron
across
948
genes.
So
in
c
elegans,
that's
a
fair
amount
of
the
genome.
A
We
computed
a
choroid
correlated
gene
expression,
score
for
each
pair
of
neurons,
so
they're,
looking
at
the
connections
between
neurons
and
that
expression,
score
and
then
they're,
looking
at
different
things
in
terms
of
gene
expression
with
signaling
and
different
types
of
signaling
between
neurons,
so
glutamatergic
and
cholinergic
signaling
are
you
know,
independent
processes
that
you
know
we
we
might
be
able
to
pull
out
of
these
data,
and
so
we
further
show
that
coupled
expression
between
hub
neurons
cannot
be
explained
by
the
neural
subtype.
A
A
That
are,
you,
know,
defining
the
sort
of
common
expression
pattern.
It's
not
just
because
they're
of
the
same
cell
type,
also
but
also
separation,
distance,
chemically
secreted,
neurotransmitter,
birth
time,
pairwise,
lineage
distance
or
topological
module
affiliation.
A
None
of
those
things
also
explain
that
the
data,
so
their
results,
suggest
that
neural
hubs
may
possess
a
distinctive
transcriptional
signature
themselves,
preserved
across
scales
and
species
that
are
related
to
the
involvement
of
hubs
and
higher
order
behaviors.
In
other
words,
these
hubs
are
sort
of
evolutionarily
conserved
and
they
serve
a
function
in
the
network,
which
is
to
organize
a
lot
of
the
connections
and
they
express
their
own
gene
expression
patterns
and
maybe
because
they're
hubs
they
do
this,
and
so
it's
a
kind
of
an
interesting
paper.
A
It's
definitely
something
we
don't
even
think
about
in
neural
networks,
but
it
you
know
it
might
be
something
important
to
understanding
like
why.
There's
this
topological
diversity,
and
so
this
paper
goes
through
it's
very
biologically
oriented,
but
it
kind
of
goes
through
the
logic
here.
Combining
the
data
sets
and
then
giving
a
result
here.
So
they
have.
A
A
Then
they
show
hub
connectivity
and
then
they
show
this
this,
so
they
show
that
this
exists.
This
is
a
well-known
result,
but
you
also
get
this
rich
club
organization
and
then
you
get
these
correlated
gene
expression,
values
so
now,
they're
looking
at
gene
expression
in
these
cells
and
they're,
looking
at
rich
club
cells
versus
feed
out
cells
versus
feed
in
cells
versus
peripheral
cells,
so
I
don't
really
know
what
those
categories
are
but
they're
in
the
paper
and
they
kind
of
have
this
these
classifications
of
cells.
A
Here
you
have
these
different
classes
of
cells,
and
you
see
there's
a
rich
to
rich
connection.
You
have
other,
like
peripheral
cells
to
rich
connections
and
those
are
all
kind
of
summarized
in
this
table.
So
the
idea
is
that
the
rich
club
is
unique
in
many
ways
and
that
there's
maybe
some
functional
reason
for
that,
and
so
so
that's
a
very
good
paper.
Again,
it's
very
biologically
oriented
and
again
here's
the
link
to
the
drive.
A
So,
if
you're
interested
in
following
up
on
any
of
these
papers,
you
can,
let
me
see
what
we
have
in
the
chat
here.
So
we
have.
This
is
this:
these
are
the
slides
for
the
ethics
abstract
from
krishna.
This
is
for
anyone
who
wants
to
look
at
the
slides
here.
We
also
have
jesse
says
thank
you.
Krishna
had
to
leave,
that's
fine
and
then
here
are
the
links.
Here's
the
link
to
this
folder
of
the
papers.
J
I
have
a
modern
moment.
Okay,
I
think
I'm
gonna
try
to
do
a
similar
thing
from
the
world
and
make
a
little
zotero
folder
for
power
painters
of
people,
because
having
a
breakfast
was.
A
Yeah
we've
been
we've
been
doing
this
with
we've
already
been.
A
With
with
endnote,
and
so
we
we
might
do
it
with
a
note,
we
might
do
it
with
zotero,
I'm
not
really
sure
yet,
but
I
think
we
already
have
one
within
note
like
we
can
talk
about
it
later,
though,.
J
Okay,
but
no,
it
was
really
good.
I
I'm
gonna
go
back
and
look
at
this
presentation,
because
that
was
a
little
interesting
and
really
the
things
that,
like
I
kind
of,
did
some
things.
J
Interesting
overlap
between
one
of
the
things
that
I
found
on
the
podcast
this
week,
and
I
mentioned
very
briefly.
I
heard
his
feedback
like
there's
a
there's
something
involved.
A
Okay,
well
thanks
any
other
comments.
A
A
A
If
you
want
to
present
again,
you
can
bring
it
to
the
meeting,
we'll
put
it
on
the
list
and
we'll
go
through
it,
and
I
you
know,
there's
so
many
different
topics
here
and
we'll
also
keep
on
top
of
the
deadlines.
So
no
immediate
deadlines,
except
for
that
waves,
special
issue-
and
we
know
about
that
so.