►
From YouTube: 2013 Fall Hackathon Q&A Session
Description
2013 Fall Hackathon Q&A Session
A
B
A
E
A
D
E
A
D
A
C
F
C
A
F
A
C
A
A
E
A
Right,
that's
the
point.
That's
the
point.
The
point
is
darker
at
the
bottom
because
of
the
light,
so
blue
is
not
blue.
Blue.
Is
this
consistency
you
perceive
in
the
world,
but
there's
no
blue
axons.
There's
no
blue,
neurons
they're!
It's
it's!
The
fact
that
that
that
that
pattern
is
is
consistent
and
persistent
over
various
motor
behaviors
like
moving
and
lighting
condition
change,
and
so
on.
A
A
C
And
then
we
say
remember
this,
but
the
data
will
not
repeat,
but
it
may
repeat
in
future,
but
we
have
to
keep
that
model
and
there
will
be
another
data
coming
in.
That's
another
model
and
we
say
remember
that
too,
and
then
we
use
two.
But
we
cannot
combine
because
the
system
will
gradually
learn
news
and
it
will
forget
the
old
one.
Well,.
A
Quickly,
right,
I'm
missing
this,
I
mean
so
I
experience
something
I
say
remember
this:
don't
forget
this!
Yes,
okay,
that's
slowly
quickly!
That's
like
saying
you
know,
make
it
permanent
right.
Yes,
okay,
we
don't
model
that
in
the
city
like
today,
I
know,
but
in
the
brain,
it's
just
there's
things
that
no
there's
some
things
that
are
well
known
about
this,
so
there
are
neuro
modulators.
A
C
C
A
Regions,
not
region
of
the
cortex,
the
subcortical
region,
which
seems
like
the
locus
corollas
and
the
what's
the
other.
These
are
very
small.
Little
emotions,
yeah
they're,
like
the
peace
size,
but
they
project
broadly,
and
you
can
read
that
so
there's
a
set
of
these
things
and
some
are
involved.
Like
a
dopamine
netherlands,
that's
involved
in
new
wars
and
anticipation
of
rewards,
so
there's
a
series
of
these
chemical
modulators.
That
say
you
know.
When
shall
I
learn
something?
When
should
I
learn
something
make
this
permanent?
Never
forget
this.
A
B
A
You
need
something
like
that
yeah,
but
it
could
be
fairly
simple.
It
doesn't
have
to
be
anywhere
near
as
complex.
What's
going
on
in
the
biological
brain
right,
you
know
it
could
be
it's
human
supervised.
We
could
say
these
are
training,
sets,
there's
lots
of
ways,
we
don't
know
yet,
but
we
don't
have
to
model
all
that
circuitry
or
complexity
right,
we're
gonna,
watch
it,
but
you'll
need
it.
A
Data
from
servers
right
yeah
and
we
assume
that's
all
good,
it's
it's
all.
You
want
to
learn
all
of
it.
There's
we're
not
trying
to
you
know,
learn
good
behaviors
and
bad
behaviors.
We're
just
saying
just
build
a
model
of
the
data,
so
we're
learning
on
everything.
It's
all
consistently
learning.
We
don't
accelerate
whether
you
turn
our
building,
just
continually
learn
what's
normal,
and
if
we
have
some
weird
behavior
and
it
happens,
then
we
start
learning
that
as
normal.
We
quickly
start
saying.
A
D
C
A
Basically,
that's
why
yesterday
we
have
other
flowers
in
your
house,
because
if
you've
never
had
any
flowers
in
your
house,
I'd
be
surprised
if
you
wouldn't
notice
it
when
you
walked
in
right.
If
you
other
times
have
flowers
in
your
house
and
she
might
not
notice
it
because
that's
normal,
because
I
want.
G
A
A
A
It
doesn't
matter
because,
because
my
expectations,
if
I
walked
into
the
indian
room
and
had
flowers
and
just
the
fact
that
it
wasn't-
and
it
was
your
model
of
a
b
and
d-
still
didn't
change
right,
you
know
one
of
my
favorite.
You
know
this.
It's
phenomenal
called
change
blindness
where
they
show
you
two
pictures.
They.
A
A
It
occupies
a
large
part
of
the
picture
and
the
other
image
the
airplane
doesn't
have
the
jet
engine
on
the
board,
so
the
jet
engines
disappear
completely
right.
When
you
show
these
two
images,
you
can't
see
the
difference,
and
so
crystal
showed
me.
This
look
at
that.
How
do
you
explain
that
I
said
both
the
airplanes
sometimes
have
jet
engines
on
the
wing,
and
sometimes
they
don't
you're.
That's
perfectly
normal.
A
If
you
show
me
one
of
those
pictures
with
the
jet
engine
above
the
wing
like
sticking
up,
I
would
immediately
jump
out
of
me
that's
wrong
immediately,
but
the
fact
that
it
showed
up
below
the
wing-
that's
normal,
that's
where
jet
engines
show
up
so
it
might
it's
not
surprised
at
all.
Sometimes
sometimes
the
airplanes
have
dead
engines
on
the
wings.
Sometimes
they
don't.
Sometimes
the
tail
doesn't
that's,
neither
one
of
those
is
a
wrong
picture
of
a
of
an
airplane,
so
the
fact
that
they
change
doesn't
matter.
A
I
just
saw
it
as
an
airplane,
but
if
there's
anything
unusual
about
that
airplane
that
never
stole
an
airplane
before
it
would
jump
right
out
so
b
and
b,
breakfast
b
rooms
often
have
flowers,
these
same
beanies
all
the
time
and
and
they
very
often
have
flowers.
So
the
fact
that
you
know
one
moment
they
had
one
moment
they
don't
you
know
it's
like
change
plans,
you
wouldn't
notice
it.
A
A
Well,
I'm
gonna,
I'm
gonna
interpret
this
in
a
certain
way
and-
and
you
may
not
notice
something
right-
you're
gonna,
you're
gonna
interpret
it
as
a
when
you
might
might
be
the
right
interpretation
of
b.
That's
just
that's
kind
of
what
you
want.
You're,
always
getting
novel
input,
which
is
never
you've
never
seen
before
the
brain.
A
Which
one
does
it
map
to
and
you're
going
to
map
it
on
to
the
one
that
you've
seen
the
most
you
say:
okay,
that's
like
this,
and
and
that's
kind
of
I
mean
I
guess,
that's
the
flip
side
of
intelligence,
that's
kind
of
what
we
want
the
system
to
do.
We
want
the
system
to
say
given
past
experiences,
as
opposed
to
saying
I
don't
know
what
this
is.
A
D
G
G
A
A
D
D
A
A
A
A
E
E
A
Well,
I
I
did
we
have
a
session
after
after
afternoon.
Oh
after
oscar,
oh,
I
said
there
was
a
thing
in
the
evening.
I
don't
think
there's
anything
planned
for
that.
I
just
have
these
three
talks
for
the
owl
conference
and
I
don't
really
know
what
the
schedule
is
beyond
that.
If
you
are,
would
you
want
to
get
together
or
something
like
that?
Yeah.