►
From YouTube: NuPIC Development Progress Review - Oct 15, 2015
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Hello,
dude
big-
this
is
matt
and
the
defense
team
in
the
office
nobody's
connected
online.
So
we're
just
going
to
jump
right
in
and
there's
a
few
things
I
wanted
to
hit
on
today,
but
I
think
we
might
as
well
do
a
quick
roadmap
review
there.
I
still
have
a
couple
of
things
to
do
to
fix
builds.
Most
of
this
is
going
to
be
put
off
until
after
HTM
challenge.
A
A
There's
still
a
lot
of
discussion
about
creating
these
api
is
computable
serializable
and
some
other
one
about
data.
So
I
think
that's
still
just
in
discussion
phases.
Anaconda
support
after
its
team
challenge.
I
really
would
like
to
look
into
that,
and
so
maybe
Scott
you
give
us
a
quick
update
on
serialization
or
other
other
related
issues,
yeah,
basically
that
progress
on
that.
So
there's
one
big
question:
mark
cream
passing
from
C++
back
to
python
and
treating
resolve
that
issue.
So
at.
A
C
A
I
think
right
now,
I'm
focused
on
the
good
old
issues,
but
once
that's
a
callback
well
organized
that
final
touch
to
get
it.
Okay.
Speaking
of
those
issues,
let's
talk
about
that.
The
main
problem
we
have
right
now
is
that
we
have
a
new
bit
core
version
that
doesn't
build
nicely
with
new
pic,
so
we're
behind
in
that
aspect,
Scott's
been
working
hard
on
it.
Maybe
you
just
give
us
a
quick
summary
of
what
the
problem
is
so
turn
tricks
with
well
we're
moving
towards
having
a
setup
where.
A
And
we'll
handle
all
of
this
was
built
in
the
seam
a
process.
So
that's
the
goal
and
it's
a
little
problematic
because
anytime,
you
change
something,
there's
a
lot
of
things
that
can
go
on
a
break
and
so
I'm
trying
to
get
as
much
done
as
possible
before
before
our
spending
all
the
time,
resolving
all
their
shoes
and
hardening
in
that
script,
and
everything
simply
try
to
do
that
at
everything
record,
so
I
think
we're
pretty
close
to
getting
practice
table,
but
right
now,
hopefully,
next
couple
of
business
aids.
A
A
B
A
A
A
A
If
we
had
more
tests
to
this
regression
sweet,
that's
something
we
should
jump
on
as
soon
as
it
fails
and
every
time
it
fails,
because
that
what
happened
last
time
was
the
email
came
and
went,
and
then
it
was
crappy
by
me.
But
everything
is
good
now
and
we
have
the
addition
of
another
test,
so
that
was
done
by
our
intern
sagen.
If
you
can
give
a
quick
overview
of
what
we
did
yeah
so.
D
Basically,
so
it's
a
new
regression
test
that
runs
the
new
big
portion
of
man.
It's
really
quite
straightforward.
It
just
ensures
that
the
scores
for
that
portion
don't
change,
so
the
test
will
fail
if
the
scores
increase
or
decrease
beyond
a
very
tight
threshold.
So
in
basically
there's
there's
no
good
reason
why
the
store
should
change.
So
if
they
do
that's
something,
we
need
to
be
aware
of
yeah.
A
A
D
A
D
A
So
we
have
a
problem
in
a
couple
of
our
repos
for
CI,
because
our
regression,
a
Russian
test,
suite,
expects
new
pic
to
be
installed
from
whatever
the
latest
iterative
build
was
that
we
pushed
to
Amazon
s3,
so
it
relies
specifically
tests
on
the
latest
iterative
build
that
doesn't
go
grab.
The
latest
version
stable
version,
we're
testing
hit
basically
head
of
master
all
the
time.
I
think
that's.
We
want
to
do
that
with
a
Russian
test.
That's
the
point,
but
the
problem
is:
we
need
to
use
some
AWS
credentials
to
get
that.
A
C
A
Bill
has
encrypted
credentials
as
environment
variables
and
you
cannot
use
those
in
PRS
because
in
your
pull
request
you
could
dump
all
the
environment
variables
as
a
part
of
the
code
change
that
you
made.
That
would
compromise
the
encrypted
values
that
should
never
be
shown
to
anyone
on
Travis,
CI
org.
So
that's
the
problem,
so
maybe
there's
a
workaround
because
I
think
it
I
think
it
was
just
for
boat.
Oh
it's
just
for
fetching
those
wheels
that
are
out
there.
A
A
A
C
A
Love
to
hear
from
him
because
he's
made
a
lot
of
progress
recently
and
we
talked
about
what
the
progress
that
Richards
made
with
widows
builts,
because
I
saw
there
was
an
a
player
job
that
was
purring
I,
don't
know
exactly
what
that
means,
but
we're
not
around
there.
Yet
we
don't
have
a
fully
working,
Dylan
I
think
we're
close
and
he's
getting
everything
set
built
successfully,
and
then
he
was
happy
or
through
this
different
hours,
and
so
I
have
a
write-up
of
the
current
status.
A
But
I
think
is
so
little
bit
of
work
left
to
do
and
again
that's
using
kind
of
custom
build
and
height
on.
So
it's
not
a
totally
generic
solution,
but
if
it's
an
option
for
running
them
will
be
an
option
for
right.
Right,
Richard's
been
really
good
about
sending
these
text
updates
every
week
or
two.
So
hopefully
we'll
see
one
of
those
and
a
good
idea
what
his
progress
is.
So
usually
we
fill
up
about
half
of
our
time.
Talking
to
David.
E
B
E
D
A
B
D
Must
be
an
Android
user
he's
a
nobody.
A
E
A
Party
all
right:
well,
maybe
maybe
you
can
join
here.
Oh
you've
been
pinging
me
here,
too,
okay,
I'm,
pasting
and
bitter
see.
Sometimes,
if
you
click
this
link,
you'll
magically
get
in
I
am
so
frustrated
with
Google
Hangouts.
They
there
now
it's
it
changes
all
the
time.
So
I
never
know.
I
have
no
confidence
that
anything
I
do
on
here
is
actually
going
to
work.
I
hear
you
so
I,
don't
know
that
your
voice
or
anything
is
actually
getting
recorded
for
the
public
here
since
you're
talking.
E
A
F
E
A
D
F
F
So
anyway,
if
you're
I
don't
know
if
you're
interested
in
what
he's
doing
but
he's
trying
to
put
in
image
data
into
the
HD
into
the
Java
HTM
and
he's
got
a
huge
kind
of
file
and
it
results
in
a
lot
of
memory
being
used,
and
so
he's
he's
working
right
now
on
reducing
the
footprint
of
that
by
using
a
bit
said
instead
of
integers.
So
you
know
so,
while
I'm
using
one
bin
studded
31
for
a
bit
so
anyway,.
C
F
That
should
that
should
help
out
things
and
right
now
he's
just
struggling
with
he's
getting
them
he's
getting
his
hand
slapped
by
coveralls.
Now,
because
you
know
his
coveralls
wants
him
to
test
putting
in
line
so
he's
like
putting
up
tests
and-
and
it's
still
telling
him
its
failings
he's
like
adding
one
test
after
another,
so
you
he's
working
that
right
now,
so
we'll
have
a
implementation
that
has
a
you
know
a
pretty
reduced
finger.
F
You
know
thumb
print,
I'll
memory,
infrared
some
kind
of
print
anyway,
so
I
started
a
blog
to
try
to
help
out
new
people
with
HTM
Java.
So
so
I
can
kind
of
learn
in
you
know
in
the
open
and
people
can
watch
me
struggle
and
relate
to
my
struggles
when,
as
I'm
trying
to
learn
HD
a
technology
and
I
put
up
some
links
there.
So
now
you
can
find
that
I'm
not
going
to
plug
it
right
now,
but
you
can
find
it.
F
You
know
in
some
of
the
conversations
we've
had
what
else
is
going
on
so
and
I've
been
working
on
my
HTM
challenge,
a
submission
so
I'm
developing
a
google
mapping
library
for
java
effects.
It's
reusing
some
couple
of
things
that
I'm
kind
of
wrapping
and
then
adding
functionality
to
so,
and
that
will
be
something
I
turn
over
to
new
mentor
after
I've
done
so
they
can
have.
F
You
know
a
self-contained
standalone
job,
a
class
that
they
can
throw
up
on
sales
calls-
or
you
know,
if
you
guys
have
tossed
something
you
can
run
it
real
easily
and
they'll.
Do
all
the
mapping
stuff
and
it'll
do
hopefully
sub
mom
that
eventually
even
even
do
prediction
of
routes
and
stuff.
But
you
know,
that's
that's
a
whole
nother
topic
anyway.
This
is
taking
me
kind
of
long
to
get
this
library
together,
but
anyway,
so
that's
going
on
what
else?
What
else?
What
else?
Well
I've
updated
in
case
I,
can't
think
of
everything.
F
F
Anti
data
no
I'm,
actually
we
were
working
on
oh
yeah,
yeah,
yeah,
so
yeah.
Well,
that
deserves
mention.
So
the
implementation
as
it
stands
had
never
really
been
tested
for
for
geo
coding
or
any
of
the
coordinate
encoders.
F
And
so
we
came
across
things
in
anti
data.
Is
a
community
member
that
is
trying
to
use
it.
I
think
for
his
own
HTM
challenge,
submission
or
some
other
worker,
another
project
he's
working
on,
and
so
he
needed
the
coordinating
code
or
two
to
work
with
the
network
API,
and
so
he
did
some
work
on
that
and
and
then
I
did
some
work
on
the
sensors
to
to
be
able
to
input
the
data
so
we've
got
then
that's
already
been
submitted
and
we
pushed
that
up
to
you
maven
now.
F
A
C
Great,
I
think
scott
mentions
the
the
progress
that
we've
made
after
the
doom
and
gloom
of
the
previous
two
weeks.
This
week's
been
really
great.
We've
got
out
of
the
804
unit
tests
in
new
pic
there's
only
ten
that
are
failing
on
windows.
Now,
oh
so
in
fact
I've
just
cleared
off
one
of
the
the
configuration
package
resource
ones.
So
we
down
to
nine
now
so
we've
got
92
trying
to
look
at
so
we've
got
the
bindings
work.
We've
answered
the
question
that
we've
been
to
an
answer.