►
From YouTube: HTM Hackers' Hangout - Nov 4, 2016
Description
HTM community event. Anyone may join, I think. This is the first time we've done a hangout on YouTube Live, not really sure how this is going to work.
A
Hello
and
welcome
to
the
HTM,
a
curse
hang
out
this
November
fourth
2016
I'm,
Matt
Taylor
and
we've
got
a
couple.
People
from
the
momenta
offices,
online,
Austin,
Marshall
and
Marcus
and
Scott
in
the
office
and
Paul
am
one
of
our
community
members.
Who's
done
a
lot
of
work
recently
on
each
team.
Related
things
is
joining
us
too.
So,
thanks
for
coming,
you
guys
and
I
have
a
short
agenda
that
I'd
like
to
push
through
I.
Think
the
first
thing
I'd
like
to
do
since
you're
here,
Scott
you
volunteered
to
do
this.
A
B
A
Okay,
so
get
a
basic
overview
of
it.
There's
still
that
forum
thread
that
just
started
a
few
months
ago
about
his
new
ideas
on
this
stuff.
So
so
that's
the
basis
of
it
and
it's
still
being
worked
out,
I,
don't
think
there's
any
significant
code
or
anything
that's
being
written
at
this
point,
it's
just
two
experiments
and
thought
thought,
experiments,
theory
etc.
Well,.
B
A
A
So
a
couple
things
I
wanted
to
talk
about
that
I
haven't
mentioned
to
the
community.
There's
been
a
lot
of
activity
internally
recently,
mostly
because
we've
shifted
our
pipelines.
So
we
had
this
initiative
to
integrate
all
of
our
continuous
integration
systems
into
one
system,
which
was
atlassian
bamboo,
and
we
partially
did
that.
So
we
ran
into
some
problems
of
and
we
wanted
to
back
off
and
just
get
into
a
green
state.
So
we
have
our
Linux
fields
running
on
bamboo,
but
we
still
have
some
issues
with
virtualization
and
windows
and
OSX.
A
As
a
repercussion
of
that,
we
have
changed
the
dependency
from
new
pic
or
two
new
pic,
whereas
before
this
change
there
was
a
direct
shot
of
pendency
from
new
pic
or
excuse
me
from
new
pic
tuna
pic
or
there
was
actually
a
dot
file.
That
specifically
said
which
Shaw
and
of
the
core
needed
to
be
installed
for
Newport
to
work
with
it.
We
remove
that
dependency
in
favor
of
specific
semantic
versioning
for
each
project.
A
So
now,
Nupur
core
has
its
own
versioning
numbers
new
pic
as
well,
and
new
pic
depends
on
a
public
release
of
nuclear
core.
So
this
encourages
us
to
release
quickly
and
at
and
agile
II.
So
that's
sort
of
the
point
we
need
to
get
something
into
new
pic
or
a
new
pic
will
will
probably
just
release
both
switch.
A
We
don't
have
any
details
about
it,
we're
still
trying
to
nail
everything
down
but
I'm,
not
giving
up
on
on
holding
community
like
live
community
events.
I
do
think
it's
useful
to
get
people
physically
in
the
same
room
that
are
talking
on
the
forearms
and
changing
ideas.
It's
been
really
nice
in
the
past
to
have
those
you
know
annual
hackathons,
just
to
bring
everyone
together,
exchange
ideas,
so
I'm
still
trying
to
make
that
happen.
A
Come
up
over
the
past
couple
months
about
that
on
the
forum
and,
of
course,
there's
you
know
Fergal
announcement
of
his
company
his
paper
that
as
that
sort
of
slant
on
it
too,
so
I
just
wanted
to
point
out.
I
think
that's
cool.
I
think,
there's
a
lot
of
potential
for
typical
machine
learning
technologies
in
HTM
to
sort
of
work
to
make
interesting
things
in
the
future,
so
nothing
but
encouraging
that
work.
But
in
there
we
missed
I
think
maybe
two
words
and
they
were
important
ones.
I
couldn't
tell
what
they
were.
A
You
said
especially,
and
then
something
cut
out
so
combining
it,
especially
with
what
reinforcement
learning
combining
HTM,
especially
with
with
reinforcement,
learning
techniques.
That's
yeah,
that's
what
I
mean
that
seems
to
be
a
trend
right
now.
I
think
reinforcement,
learning
is
this
kind
of
hot
and
it
it
mixes
well
with
HTM
in
a
way
because
they're,
all
there
I
think
RL
has
more
capability
for
for
temporal
types
of
things.
A
C
That's
that
after
yeah,
that's
that's
the
trickiest
part
is
how
do
you
know
and
if
you
look
at
the
thread
that
I
put
on
there,
one
of
the
steps
is
imagined
possible
future
actions.
That's
that's
very
unbiased
and
okay
I'm
going
to
try
this
you
what
that
would
be
try
something
else
see
what
that
would
be.
That's
not
really
up.
You
know
the
way
it
would
work
in
reality.
Yeah.
A
Yeah,
which
is
fine
for
experiments
and
as
you
know,
you
know
you
mented.
Our
research
is
constrained
by
biology,
but
there's
nothing
that
says
that
the
community
must
be
constrained
by
biology.
So
you
can
take
this
theory
and
do
what
you
like
with
it
and
I
so
I
think
that's
cool,
I'm
glad
to
see
those
types
of
experiments
being
done.
A
A
This
is
a
going
to
be
short
and
it's
mostly
because
David
Ray
is
not
here,
I'm
sure
he's
watching
now,
often
ok,
so
you
know
you
should
show
up
here
and
few
seconds
all
right,
I
hope
to
see
like
you
should
give
him
like
one
minute,
but
ok.
I
guess
that
is
it
for
the
HTM
ackers
hangout
will
we'll
do
this
again
in
about
a
month,
and
hopefully,
I'll
have
at
least
1
HTM
school
published
by
then
that's
all
I
got
thanks
for
joining
everybody
thanks
thanks
thanks.