►
From YouTube: S301 - AI for Every Developer - Gian Paolo Santopaolo
Description
Artificial Intelligence is changing everything from how we live to how we work and even how we think. As a developer, you should be part of it. And this is the right place where to start with a deep overview of what AI is, where it is going, and how you can be a part of it.
A
Hi
everyone
thank
you
and
good
morning,
good
evening
or
good
afternoon,
wherever
you
are
in
the
world
and
welcome
to
artificial
intelligence
from
for
every
developer,
I'm,
John
Paolo
and
the
work
I'm
doing
here
at
ibv.
It's
a
lot
of
work
about
emerging
technologies,
and
this
involves
artificial
intelligence.
Material
intelligence
is
a
field
that,
since
some
years
ago,
it
was
restricted
to
a
bandwidth
of
some
some
scientists
with
the
work
Microsoft
and
other
big
companies
in
the
movement
they
are
making
in
artificial
intelligence,
it's
not
possible
to
get
artificial
intelligence
and
develop
artificial
intelligence.
A
Every
US
can
develop
artificial
intelligence.
So,
let's
dig
into
the
slide-
and
let's
start
talking
about
this
course,
because
we
have
a
lot
to
talk
and
not
that
much
time
so
about
this
course.
When
you,
the
take
away
you
will
get,
is
you
will
understand
at
the
end
of
the
she
will
understand
what
our
intelligence
is
so
define
it
in
the
context
of
science?
What
machine
learning
is
and
also
what
the
learning
is?
The
way
you
actually
can
develop
artificial
intelligences?
The
way
you
can
infuse
artificial
intelligence
in
your
application
to
make
them
smart?
A
We
will
you
will
also
understand
the
core
concept
of
deep
learning.
Some
of
the
technology
is
some
of
the
terminology
and
the
algorithm,
and
we
will
also
develop
some
cool
stuff.
So
let's
go
ahead:
artificial
intelligence,
it's
something
that
it's
it's
there
since
long
time
ago,
since
early
forties
in.
A
1959,
the
concept
of
machine
learning
was
was
introduced
for
the
first
time,
but
what
is
literally
changing
the
world
is
the
terms
of
is
the
introduction
of
deep
learning.
Deep
learning
was
introduced
before
2012,
but
actually
the
first
time
it
was
used
to
train
a
computer
to
make
a
computer
intelligence.
It
was
2012
in
2012,
and
this
was
reported
in
the
mayor
journal
in
the
United
States.
A
This
is
something
that
maybe
not
everyone
realize,
but
it's
something
very
important
because,
let's
say,
let's
start
from
1995
in
1995
we
got
five
million
transistors
on
a
chip
and
it
was
the
population
of
New
York
City.
At
that
time,
1995
it
was
the
launch
of
the
first
inter
Pentium
with
Windows
95
was
also
lunch.
Luncheon
and
Henry
was
changing
the
world
2005
the
number
of
transistors
on
a
single
chip
growth
to
160
millions.
A
This
is
the
population
of
the
entire
East
Coast,
and
it
was
the
era
when
the
heading
for
was
coming
out.
Internet
was
growing
and
there
was
the
the
mobile
team
was
near
the
corner
2010.
We
got
1
billion
transistor,
and
this
is
the
post
iPhone
post
iPod
connected
world,
and
we
were
doing
gestures
and
voice
recognition
in
our
living
room
with
150
bucks
2015.
We
got
7.5
billion
transistors
on
a
chip
and
that's
a
and
that's
a
transistor
for
every
single
man,
woman
and
child
on
earth.
A
How
big
it
is
how
big
this
drawing
is
going
on
3
years,
whether
2017
we
fill
it
up
another
planet.
It
took
30
years
to
fill
up
to
New,
York,
City
and
just
ten
years
to
fill
up
the
entire
world.
That's
the
computer
power,
the
compute
power.
That
is
changing
the
face.
How
we
will
see
the
world
in
the
next
couple
of
years
and
that's
why
now
is
the
right
moment
to
jump
in
in
the
artificial
intelligence
field
and
start
developing
for
for
it,
so
artificial
intelligence,
artificial
intelligence,
it's
a
branch
of
science.
A
It's
the
science
of
making
things
smart.
It
can
be
defined
as
human
intelligence
exhibited
by
machine,
and
this
is
this-
is
in
general,
artificial
intelligence,
the
stutter
d'art.
What
we
can
do
today
with
artificial
intelligence,
artificial
intelligence
today,
is
like
something
done.
Don't
you
think
the
artificial
intelligence
of
Terminator?
We
are
not
there.
We
won't
be
there
intelligence.
Today,
it's
a
form
of
narrow
AI.
A
A
system
can
be
just
one
or
fuel
the
finer
things
as
well
or
better
than
a
human,
and
this
is
important,
so
it
can
recognize
something
or
maybe
detect
a
credit
card.
Fraud
in
there
are
several
use
cases
in
artificial
intelligence.
Like
object,
recognition,
speech
creatively
like
stand
transfer,
as
you
see
in
this
picture
on
the
left,
the
original
picture
in
the
middle
of
the
style
that
will
be
transferred
to
the
original
picture
and
on
the
right,
you
will
see
there
the
result.
A
A
Machine
learning
involve
teaching
the
computer
to
recognize
patterns
by
example,
rather
rather
than
programming
it
with
with
new
specific
rules,
but
we
like
we
are
usual
for
doing
it.
If
they're
else,
no
much
ma'am,
surely
something
different.
So
we
will
teach
it
to
recognize
a
pattern
and
this
pattern
can
be
found
within
data
within
data
and
basically
is
creating
algorithm
that
can
learn
complex
function
from
data
and
make
prediction
now
out
of
them.
A
So
as
we
as
we
said,
machine
learning
is
predicting
stuff
and
it's
intelligent
because
it
takes
some
data
to
train.
The
system
learns
pattern
from
this
data
and
then
classifier
classifies
new
data.
It
has
never
seen
before
it
could
this
later
classified
it.
The
way
that
the
result
is
gave
out
is
giving
us
like
a
best
guess
of
probability,
so
the
difference
between
traditional
programming.
If
we
can
take
an
example,
maybe
from
spam
filter
and
if
you
were
asking
to
develop
a
spam
filter
today
with
classical
development.
Let's
say
it's
a
chef
development.
A
What
what
you
are
going
to
do,
you're,
making
a
lost
long
list
of
if
and
then
so,
a
long
list
of
rules
with
machine
learning?
We
can
train
a
computer
system
to
classify
some
email,
and
then
this
system
is
seven
once
its
trainer
with
a
lot
of
data
is
able
to
apply
the
learning
model
and
classified
the
email.
In
this
case,
deep
learning
can
be
journal
defined
as
a
technique
of
implementing
machine
learning.
One
such
technique
is
a
concept
known
as
deep
neural
network.
A
Let's
say
deep
learning
is
the
context
of
deep
neural
net
and
deep
neural
network
is
where
our
code,
our
artificial
intelligence
code,
is,
is
written
on
so
the
ways
of
developing
artificial
intelligence.
We
have
several
ways
actually
three
ways
of
developing
artificial
intelligence:
the
first
one
the
easy
way
using
artificial
intelligence
as
a
service.
A
That
is
just
using
a
service,
so
we
just
need
to
invoke
our
our
services.
In
these
cases
the
sample
is
a
settlement.
Analysis,
section
parameter
our
subscription
emulator
subscription
key
and
then
send
the
send
a
request
to
the
to
the
service,
and
we
will
get
a
result,
nothing
more
simple
than
that
more
simple
than
this
same
wave
still.
Is
it
but
some
more
work
and
also
the
bill
2018.
We
have
custom
cognitive
services.
This
means
that
we
are
given
the
ability
to
customize
general
services
to
our
specific
domain.
A
A
Just
go
to
the
custom
vision
site,
create
a
project
if
you
want,
if
you
want
to,
if
we
want
to
make
the
classification
and
the
project
type,
should
be
classification,
classification
type,
not
a
class,
because
we
want
to
classify
several
the
demands,
different
elements,
and
if
we
choose
us
domain
any-any,
that
is
compact,
we
can
then
export
it
to
a
lot
of
to
a
lot
of
stuff.
That
I
will
show
you
later
on.
A
So
once
we
have
created
the
project,
just
drop
your
figures
in
this
case
they're
just
some
fishes
some
flowers
and
some
sticky
figures.
Let's
train
our
model,
we
get
the
result.
If
we
are
happy
with
the
result
of
the
precision
of
the
train
and
machine
learning,
we
can.
We
are
good
to
go
and
we
can
export
as
I
showed
before
we
made
that
compact
and
in
the
project,
and
then
we
can
export
it
to
an
annex.
A
What
does
it
means?
It
means
that
then
we
can
use
it,
for
example,
in
our
uvp
application
and
to
use
it
again,
it's
very
simple,
just
below
the
model
put
it
in
a
work
put
it
in
our
project
and
the
project
will
automatically
generate
all
the
classroom
called
need
to
vote
to
vote
the
almanacs
model,
so
we
have
just
to
load
a
model
and
then
in
this
case
it
was
just
for
recognizing
the
shapes.
So
it's
like
we
are
reading.
A
Okay,
how?
How
is
possible
that
we
can
run
artificial
intelligence
actually
in
our
in
our
client?
Whatever
it
can
be,
can
be
a
server,
it
can
be
a
works.
Everything
can
be
our
our
new
VP
application.
It
tends
to
witness
and
he'll
witness.
Nl
is
another
piece
of
technology
developed
by
magic
that
Alps
us
using
artificial
intelligence
without
the
need
of
deep
going
deep
in
artificial
in
the
core
stuff,
for
example,
with
the
artificial
intelligence.
A
We
can
focus
on
our
on
our
domain
so
and
what
this
means
that
we
can
use
our
model
without
thinking
how
to
make
it
works
to
make
it
works
issue
the
the
model
when
it
gets
ugly,
but
it
is
a
skill
for
a
prediction:
it
can
be
that
it
has
to
run
against
the
CPU
or
against
the
GPU
with
women,
and
all
these
stuff
are
done
under
the
hood.
We
don't
have
to
care
about
it.
A
At
the
end
of
the
course,
I
was
like,
with
all
the
links
that
I
collected
during
these
years,
that
you
can.
You
can
get
use
it.
So
there
are
also
list
of
a
lot
so
already
training
modem
that
you
that
you
can
use
so
as
I
say,
also
they're,
all
the
things
related
to
the
other.
We
don't
have
to
care
about
it.
Everything
is
kept
by
Windows
ml,
so
that's
cool
as
to
Microsoft,
also
for
this
great
piece
of
technology
that
we
can
use.
A
If
you
want
to
get
started
with
our
Analects,
just
not
very
site
and
go
to
see
their
getting
started
in
their
tutorials,
because
one
of
the
other
important
feature
of
onyx
that
you
can
experience
is
that
around
the
world.
If
you
look
for
models
or
machine
learning,
deepen
your
network
models
or
up
to
develop
it,
you
should
have
developers
with
several
different
frameworks
like
tens
of
coffee
or
something
else
with
onyx
with
the
project
onyx
with.
There
is
also
the
way
for
exporting
project
for
made
with
other
SDKs
to
onyx.
A
If
you
want
to
get
zoom,
if
you
want
to
get
this,
you
can
also
train
your
data
on
the
cloud.
You
can
also
use
them
ml
tools,
minimal
tools
and,
as
I
say
view
or
television,
a
eyesight
okay,
and
these
are
the
the
first
two
ways
of
using
AI,
but
we
are
not
really
developing
a
we
are
just
using
it,
and
so,
if
you
want
to
dig
deeper
and
get
our
hands
dirty
right,
our
artificial
intelligence
code,
we
have
to
go
through
some
process.
A
First
of
all,
we
have
to
understand
the
core
concepts
terminology,
an
average
of
artificial
intelligence,
so
that
then
we
can
develop
our
first
determining
application
so
again,
which
is
the
way
to
develop
an
artificial
intelligence.
We
need
the
second
data.
Actually,
we
need
a
lot
of
data
to
build
and
train
the
model,
and
then
we
can
we
can
deploy
it.
We
can
usually
we
can
use
it
and
we
can
deploy.
It
was
with
oil
and
eggs.
A
Let's
go
a
step
back
to
build
the
model
and
to
train
it
before
someone
said
Oh
before
women
before
ml,
what
threat
was
was
developers
was
released
before
the
visual
studio
tools
for
AI
was
released
as
a.net
developer.
The
way
to
develop
artificial
intelligence
is,
it
was
to
go
totally
out
of
our
comfort
zone.
We
need
to
use
Python.
We
need
to
use
tools
that
we
are
not
used
to
thanks
to
the
mi
tools,
to
the
IAR
tools
for
visual
studio.
A
We
are
able
to
run
and
build
artificial
intelligence
using
the
truth
set
using
visual
studio.
Actually
that
concept-
and
there
is
also
an
AI
tools
for
visual
studio
code,
so
using
the
tools
that
we
can
define
our
usual
to
use
nowadays
everyday
developer,
then
to
train
the
model.
We
can
use
our
local
hardware
or
we
can
train
it
on
some
custom
dedicated
mr.
machine,
an
answer
with
a
lot
of
computational
power
related
to
a
a
I
know.
So
that
means
lots
of
gpgpu
bother.
A
So
let's
get
a
step
back
prepare
today
what
what
it
means
prepare
the
data
profess
the
data.
It
means
first
to
identify
the
actual
attributes
of
the
things
you
are
trying
to
classify
if
I
am
trying
to
classify
fruit,
maybe
the
features
and
the
attributes
that
I
need
are
color.
Weight
dimension
usually
refers
to
the
number
of
attributes
on
the
case
of
cholera,
any
weight
we
have.
We
have
an
attribute
with
two
dimension
so
feature
or
absolute
is
one
particular
table
data.
That
generally
is
called
data
points.
A
A
A
Regarding
data,
depending
on
the
problem
you
are
trying
to
solve,
while
you
are
trying
to
infuse
AI
data
can
be
anything
can
be
database
roles
can
be
Sun,
st.
pölten,
be
video,
simple
can
be
images
can
be
test
anything
you
need
now.
There
is
some
important
to
say
the
data
challenge
to
find
the
red
data
for
solving
your
problem
is
important
when
you
are
actually
developing
when
you
are
actually
trying
to
solve
your
problem.
A
If
you
want
to
learn
artificial
intelligence,
you
don't
have
to
care
at
all
about
it,
because
there
are
a
lot
of
already
plainly
datasets
around
and
again
in
the
last
one
of
the
last
slide.
I
have
there
is
a
lot
of
links
and
there
is
covering
also
this
part
how
to
get
data
so
that
you
can
start
learning
artificial
intelligence,
but
anyways
that
is
really
important.
When
you
are
then
going
into
production,
they
were
looking.
You
are
really
solving
your
problem.
A
A
A
One
of
the
points
with
Archie
boots
named
number
of
lights,
:
wait.
So
the
two
data
points
that
we
are
going
to
train
is
just
dog
for
legs
color
like
ten
pills
and
she
can
chew.
X
orange
five
kilos.
Now,
if
we
train
the
model
just
with
tune
these
two
data's,
and
then
we
ask
the
model
for
to
evaluate
a
cow
with
four
legs,
core
break
and
200
kilos.
I
believe
that
the
model
will
predict
dogs
because
it
only
knows
about
dogs
and
key
in
picture
so
and
chicken.
A
So
this
is
the
best
guess
there
is
a
nice
sample.
That's
sample
that
one
of
my
fellow
friend
from
the
IIT
was
telling
me
and
she
has
a
daughter
and
they
were
useful
with
dog
because
they
have
a
dog
at
home.
Then
they
move
it.
She
actually
was
joining
Microsoft
and
there
was
a
lots
of
horses
around
and
her
daughter
come
to
him
and
say:
maybe
that
did
not
be
here.
There
are
some
big
dogs,
that's
that's
funny
because
she
never
sold
a
horse,
so
the
father
has
to
tell
her
no
darling.
A
This
is
not
the
dog.
This
is
not
yours,
so
that
then
the
brain
is
trained
and
he
also
understand
how
to
recognize
and
differentiate
dogs
from
horses.
So
there
are
some
no
way
to
train
a
model.
Usually
they
are
divided
in
supervised,
learning
and
supervised,
learning,
supervised
and
unsupervised,
learning
supervised
living
and
when
they
Marshall
England's
training
with
data
in
this
data
Allah
blood.
What
is
means
a
blood
in
our
case
where
what
we
want
to
classify
animals,
the
wagon
will
be
the
range
of
the
young
dog.
A
And
then
we
give
three
inputs,
number
of
lights,
:
wait.
We
are
telling
the
system
what
not
liable
we
do
expect.
We
will
Emily
this
data
to
predict
future
and
sonata,
so
what
it
means.
It
means
that
we
can
really.
We
have
a
boundary
we
can
represent.
If
we
can
represent
it
in
two
axis
dimension,
we
will
have
a
boundary.
The
boundary
can
be
a
straight
line,
a
cool
or
whenever
a
function,
let's
say,
and
if
the
libelous
dog
should
be
a
circle.
A
So
we
were
saying
level
data,
and
then
we
have
unsupervised,
learning
and
well
visit.
Learning
is
where,
when
we
are
developing
artificial
intelligence
and
machine
learning
lives
from
a
data
set,
so
we
have
imagined
we
have
some
points
on
a
graph
representing
three
different
things,
and
the
machine
learning
system
must
recognize
itself
that
there
are
three
different
clusters
and
it
has
to
categorize
itself.
A
This
is
tricky
because
the
number
of
clusters
may
not
be
known
in
advice,
so
it
has
to
take
the
best
guess
sometimes
also
the
cluster
has
not
clear
as
the
one
we
have
in
the
picture
below,
so
the
best
guess
can
be
also
not
not
good,
and
so,
when
we
are
talking
about
unsupervised
learning.
Also,
another
concept
is
the
way
of
training
that
and
it
can
be
reinforcement,
learning
so
learning
by
trial
and
error
rewarding
and
punishing
the
system.
What
this
means.
A
This
means
that
machine
learning
in
the
case
of
video
games,
teaching
yourself
how
to
play
a
video
games,
learn
by
playing
the
games
millions
of
time
and
the
system
is
the
word
when
it
makes
a
good
move
when
it
losses.
We
give
him
no
or
negative
reward
over
time
over
millions,
billions,
iteration
Marshall
learns.
Machine
learning
learns
how
to
maximize
rewards
without
the
even
explicitly
tell
telling
them
the
rules
it
can
lead
to
better
the
human
performance
when
it
finds
the
part
that
no
one
taught
to
doing
before,
and
this
is
really
really
amazing.
A
So
there
are
many
ways,
a
machine
learning
to
learn
pattern
to
classify
that,
as
we
say
in
this
example,
videos
aligned
to
the
v2
divide
requested
when
we
can
predict
future
that
saying
anything
above
the
line
Islam
by
a
cluster.
Anything
anything
below
the
line
is
owned
by
another
cluster,
but
we
can
also
use
a
cubic
curve,
as
we
saw
before
so
instead
of
a
straight
line,
we
can
have
a
cubic
curve.
As
you
see
in
the
picture
below
and
the
way
the
machine
learning
will
predict
future
point.
A
So
now,
let's
talk
about
neural
networks.
Our
main
consists
of
something
like
86
billions.
Interconnection
of
neuron
engineering,
respond
to
a
certain
stimuli
and
passes
output
to
another.
There
may
be
imagine
them
dedicated
to
recognize
God
some
for
any
attributes.
We
want
each
having
a
different
weight.
A
The
weight
is
important
that
the
feature
is
to
the
to
develop,
recruiting
law
of
understanding
animals,
for
example,
when
the
neuron
fires.
If
all
these
neuron
files,
your
brains,
tell
you
you
so
attacked
in
an
animal
Network,
so
a
model
that
is
loosely
modeled
like
the
brain
neural
network
can
use
it
to
calculate
the
probabilities
for
featured
they
are
trained
to
look
at
for
so
with
neural
networks.
We
try
to
mimic
the
functionalities
of
a
brain.
Do
you
remember
when
I
told
you
the
story
of
the
horse
and
the
dog?
That's
why?
But
that's
really!
A
So
this
is
the
representation
of
an
artificial
neural.
So
we
have
three
hundreds
on
the
Left
corresponding
to
the
inputs.
Come
to
the
network,
let's
say:
values
on
7.1
7.6
1.4
and
these
are
the
weight,
are
singing
assignment
to
the
corresponding
mean
good.
So
this
is
the
input
in
the
way
it
would
get
multiplied
by
dre
they
arrested
weight
in
their
son
and
the
son
is
taken.
A
So
if
we
consider
in
a
train-
but
we
have
X
1
X
2
X
3-
and
if
we
consider
the
3
weight,
it
can
be
W,
1,
W,
2
and
W
3.
We
have
done
the
son
of
it
and
after
we
we
Surya,
we
usually
have
the
baby
s
and
we
also
seen
the
coldest
to
this
room
to
get
the
Sun
and
the
bit
bias
is
just
a
custom
number,
let's
say
1,
which
is
added
for
scaling
purposes.
A
So
the
Nissan
will
be
something
like
this
one,
it's
not
necessary
to
add
bias,
but
it's
a
good
products
and
I
see,
and
it's
that's
good
for
speeding
up
the
process.
After
adding
the
ballasts,
we
reach
the
threshold
step
and
if
the
new
Sun
calculated
is
above
the
threshold
value,
then
Evelyn
get
excited.
So
it
means
it
passes
the
value
and
it
passes
up
to
the
output,
otherwise
it
doesn't
so
exact
in
goes
there,
and
then
it
goes
on
the
other
side.
Now,
with
this
concept,
this
is
the
math
behind
or
this
severe
developer.
A
We
are
not
really
caring
about
that,
but
also
in
my
my
needs.
I
have
good
a
good
book
where
to
start
from,
and
the
book
is
actually
made
out
of
professor
of
the
California
University,
and
when
you
get
this
book,
you
get
also
access
to
the
full
semester
recordings.
So
it's
a
video
recording
where
the
professor
is
teaching
the
course
buses
buses
on
that
book.
That's
very
good!
If
you
want
to
dig
deep
into
the
to
the
matter
of
this,
the
activation
function
that
we
saw
over
there.
A
There
is
another
dis,
another
part
of
the
process
of
Neverland,
and
you
should
the
nerve
cell
activation
faction
function.
Usually
they
are
used.
The
most
usual
90%
of
the
time
is
the
rebel
one
or
the
a
page
or
the
sigmoid.
The
attend
age
goes
from
use,
one
to
one,
the
sigmoid
most
from
zero
to
one.
These
are
the
most
used,
and
there
is.
There
is
something
to
say
about
this:
about
artificial
intelligence,
official
intelligence.
It's
so
complex
that
it's
not.
A
It's
not
an
exact
science.
The
way
we
develop
artificial
intelligence
is
by
testing
so
because
of
this,
we
can
apply
to
our
Marilyn
one
activation
function
or
another.
When
so
that,
then
we
can
trim
the
model
with
an
activation
function,
get
the
result,
ran
the
model
with
another
activation
function
and
get
the
result,
as
well
as
using
different,
deep
neural
network
patterns
right
by
testing
the
data
by
testing
the
data
and
getting
the
result,
we
will
get
the
most
important,
the
most
important
part,
so
less
I
want
to
show
you
something
about
that.
A
A
So
then,
is
that
the
basis
that
is
a
database
of
M
brittle
digits
and
we
have
seven
70,000
bubble
at
the
example
and
10,000
samples
that
can
be
used
to
get
the
result.
This
is
another
important
concept
about
artificial
intelligence.
When
we
say
when
we,
when
we
train
our
model,
then
if
you
want
to
test
it,
we
have
to
use
a
data
set
that
is
never
shown
before,
because
if
you
use
the
same,
that
is
max
volume
right.
A
It's
not
good
to
use
the
same
data
that
we
use
it
to
change
the
system
for
them
testing
it
out.
Okay
right,
so
this
is
another
another
important
thing
always
to
divide
training
data
set
from
test
data
set.
So
here
how
what
you
like
you
can
see,
there
are
several
different
classified
classified
users
to
create
machine
learning
model.
So
what
does
it
means?
We
get
the
data
set,
we
trained
model
with
one
of
these
classifier
and
also
with
post-processing,
and
then
we
get
the
result.
A
A
So,
okay,
we
got
the
concept
of
an
arrow,
then
imagine
that
we
can
stock
a
multi
right
now,
like
it
happens
to
the
brain
in
all
stacking
up
playing,
Activision,
Maryland,
a
trust
to
create
intelligent
system,
intelligent
systems.
Why?
Because
we
are
talking
about
deep
neural
network
and
deep
neural
network.
Is
silicon
system
interconnected
in
layers
of
nail
on
neurons?
There
are
also
submit
the
wire
between
the
able
to
the
output.
A
Initializer
can
learn
from
the
one
before
this
is
important
if
the
buyer
can
usually
are
usually
a
lower
dimension
so
that
they
can
generalize
and
not
over
50
put
data
middle
layer
can
learn,
features
of
features
and
a
simple
example.
It's
like
first
wire.
It's
like
lead
to
lead
to
understand.
That
is
a
face
part
and
then
the
other
wire
leads
to
faces
and
so
far
solve
the
most
tune.
A
Yes,
some
some
machine
learning
types
we
have
relation,
that
is
to
predict
numerical
values
like
price
of
the
house
classification.
So
if
this
cut
human
or
whatever
last
animal
muscle
or
another
example,
so,
for
example,
the
relative
products
of
enhancer
and
then
we
have
a
sequence
prediction.
So
if
we
have
a
sequence,
1
2,
3,
4
5,
we
want
to
predict
the
next
number,
so
we
are
starting
our
own
application
so
to
develop
our
application,
the
tools
or
real
important
Visual,
Studio
Tools.
You
get
go
ahead
and
go
to
grab
it
and
install.
A
There
is
also
a
simple
repository
on
channel
9.
That
is
a
very
good
video
talking
about
visual
skill
to
how
to
install
and
how
to
use
it.
We
know
well
in
a
man,
it's
something
really
important.
It
was
a
relation
something
like
6
months
ago.
I
think
I
build
an
asset
of
field,
and
before
this
shoe
guys,
as
I
told
you
the
only
way
to
develop
artificial
intelligence,
it
was
just
to
go
out
of
our
comfort
zone.
Now.
A
What
we
can
do
is
using
the
visual
you
know,
plus
for
a
high
and
also
we
never
have
between
a
man.
What
we
can
do
is
to
develop
artificial
intelligence
in
seizure
now
and
I
was
really
and
I
will
show
you
what
what?
What
is
the
defense
we
are
talking
about,
modified
National,
Institute
then
started
and
Technology
database
of
I
am
freed
and
Egypt
and
we
are
gonna
use
it
for
the
for
the
rest
of
the
course.
This
is
an
important.
A
This
is
an
important
one,
so
let's
develop
our
first
application,
so,
okay,
so
I
was
talking.
I
was
talking
about
the
enemies
database
and
so
then
lemon
is
the
base.
This
is
another
thing
that
you
will
find
whether
my
resource,
it's
as
I
told
you,
is
a
database
of
written
digit
database
in
in
consist
of
seven
total
70,000
in
major
written
visit,
major
28
pixel
by
28
pixel.
Now
what
I
prefer
think
that
we
have
to
do
when
we
have
when
we
will
have
sets
of
data
that
is
multi-dimensional
and
usually
multi-dimensional
data?
A
This
is
another
when
I
think
to
understand
is
that
we
need
to
get
this
picture
and
load
in
a
single
dimensional
array
for
doing
that,
we
are
doing
what
is
called
flattering,
so
we
are
going
to
take
so
it's
like,
as
we
fade
it's
28
pixel
by
28
pixel.
So
we
get
the
first
row
28
pixel
and
we
will
start
putting
in
in
a
row.
A
Then
we
get
the
second
row
and
we
go
and
we
put
it
on
the
side
on
the
side
by
adding
if
you
multiply
28
by
28,
if
comes
out
1770s
184
pixel-
and
this
is
the
our
our
vector
that
we
are
gonna
use
for
for
such
kind
of
of
stuff.
Now
why
I
won't
use
well?
Why
I,
like
the
enemy's
database
I
like
the
Emily's
database,
because
it's
something
visual?
So
when
we
train
when
we
train
the
model
we
get
the
result,
we
get
a
result
visually.
A
We
understand
it,
the
marshal
invincible
friend
or
not,
because
because
the
data,
the
data
visuals.
So
if
we
have
right
one
for
example,
then
we
ask
the
machine
learning
to
get
it.
We
can
understand
if
it's
good
or
not,
so
there
is
another
another
another,
very
nice
resources
that
I
don't
see.
Let
me
let
me
find
because
it's
something
that
I
want
really
to
to
show
you
so
bookmarking
tools,
you
bookmark
sidebars.
A
A
So
this
is
a
paper
piece
this
this
guy
is
resist
read
about,
and
he
was.
The
paper
was
based
on
the
enemies
database.
So
this
is.
This
is
the
result
of
Estonia
of
a
semester
or
so,
and
one
thing
that
I
was
telling
you
about
the
data.
Is
that
but
okay,
let's
see
if
we
can
visualize
it,
because
it's
very
nice,
so
it's
getting
a
bit
longer
to
run
it
but
this
morning,
so
why
why
it's
good
to
you
visual
data
when
we
are
starting
understanding,
we're
starting
understanding,
artificial
intelligence?
A
So
here
you
can
say
any
of
this
point
is
the
point
user
to
train
the
model
and
what
we
can
experience
over
there.
This
is
an
aid,
but
this
is
in
the
in
the
context
over
over
made
okay.
So
this
is
an
X
in
the
contest
over
six
right.
Why?
Because
it
really
seems,
since
we
are
not
sure
if
it's
an
eight
or
six-
and
this
is
where
the
artificial
intelligence
may
become
can-can
filed
right.
This
is
a
five,
but
it
seems
yeah
right,
you
see,
it
can
be,
it
can
be
and
solve
so.
A
I
can
I
split
like
this
now
I
cannot
find
the
source
on
my
on
my
laptop,
but
you
can
go
online
and
also
in
the
list
of
the
sources.
I
will
give
you
the
results
of
the
link
for
the
sample,
our
realities
for
the
sample
re
master.
So
this
is
the
feminist
database
implementation
using
actually
the
CNT
K
SDK.
A
So-
and
this
is
this
is
Python,
so
we
are
already
in
our
comfort
zone
because
we
are
using
our
tool
right,
but
this
is
Python.
So
it's
a
step
ahead
because
we
don't
have
to
use
other
tools
so
kudos
to
the
visual
studio,
I
tools
for
the
with
for
the
team.
These
tools
we
are
still
out
of
our
comfort
zone.
Nevertheless,
I
want
to
show
you.
This
is
the
most
important
part
of
the
code
where
they
are
creating
them.
A
A
Okay
with
the
breakpoint
yeah,
we
can
also
use
all
the
features
a
permissions
to
it,
that's
fantastic,
so
it's
creating
the
model.
We
are
adding
some
Liars
to
our
tour,
deepening
room.
That's
what
you
remember
the
slides
where
we
was
talking
about
when
we
was
talking
about
that
the
things
the
the
lives
from
the
deep
neural
network.
This
is
creating
the
wires
for
for
the
further
model
and
then
the
get
the
data
gets
downloaded
by
every
local
software
is
not
going
to
download
it
and
then
exactly,
let's
go
there
and
then.
A
A
A
A
Okay,
we
can
open
file
explorer
if
my
laptop
is
able
to
do
it
and
it's
been
slower
down
here
we
go.
This
is
the
model.
The
signal
TK
are
ready.
This
is
the
last
line
of
the
code.
Let's
see
so
there
we
are
running
out
of
time.
I
want
to
show
you
see
sharp
samples,
okay
during
smaller,
so
that
then
we
can
use
the
onyx
model
another
component,
so
let's
say
we
are
going.
It
was
a
bit.
We
are
a
bit
late.
A
So
what
something
that
I
want
to
show
you
is
how
to
do
it
with
minimal,
with
minimal
just
create
just
create
a
project
and
the
dependency
to
the
the
dependency
to
the
windmill
package,
and
then,
for
example,
here
am
using
the
iris
data
this.
This
is
another
another
data
set
for
classifying
for
classifying
iris
flower
passes
on
this
on
this
part.
So
how
long
it
is
it's
yet
classified.
A
So
here
we
go,
we
can
do
it,
we
can
do
it
on
c-sharp,
just
create
a
pipeline
load,
the
data,
the
same
code
that
we
was
looking
at.
They
see
MDK
before
now.
We
are
doing
it
in
c-sharp.
We
can
add
our
label,
we
can
add
our
features
and
then
this
is
the
most
important
part.
We
can
add
our
classifier
classify
the
Braille,
and
when
we
have
the
classifier,
then
we
can
also
change
the
data
and
predict
it.
A
I
have
done
also
another
one
sample
that
I
want
to
show
you,
and
also
this
repository
is
part
of
array
available.
It's
also
the
link
at
the
end
of
slide.
It's
in
my
procedure.
So
again
this
is
another.
This
is
a
bit
more
complex,
like
example.
This
is
what
I
want
to
show
you,
because
this
is
also
you
see
using
regression,
so
so
fun
training
the
data
testing,
the
data.
We
can
read
it
and
then
we
can.
We
can
train
our
model,
we
can
evolve
well,
there
is
thought
and
then
use
our
values,
our
models.
A
A
A
A
But
I
will
update
also
this
repetition
with
also
this
sample
and
then
what
what
we
have
to
do
is
just
drop
the
when
I
next
package
run
an
x5
over
there
when
you
drop
over
there
it
the
visuals
will
be
already
create
all
the
classes
you
need
to
use
the
classifier,
then
you
are
good
to
go
and
use
it
so,
and
this
is
the
part
right
as
I'll
show
you
in
the
code.
This
is
the
the
model,
but
keep
in
mind.
This
is
the
model
that
we
develop
it
now.
So
this
is
the
real
difference.
A
So
let's
say
you
can
you
can
look
at
the
color
by
yourself,
so
it's
loading,
the
model
I
can
write
something
over
there
and,
let's
say
recognize,
and
that's
it.
If
you
want
to
say
the
process
is
really
simple.
Let's
say:
let's
do
it
like
this
erased,
let's
make
a
cell
regular
so
recognize
what
I'm
doing
over
there
I'm
getting
the
in
the
stroke
from
the
novice
confirm
everything
in
the
way
the
model
inputs
likes,
the
impotent
passing
to
the
model.
I
am
asking
to
evaluate
it
when
it's
elevated
I
get
I
get
the
result.
A
From
this
case,
it's
a
bit
complicated,
so
I
get
an
array
with
all
the
guests
and
I
understand,
which
is
the
best
guess,
I'm
just
iterating
the
Iranian
getting
the
best
guess
that
in
this
case
is
say,
say
in
this
case:
it's
a
six
okay,
but
this
time
and
the
model
wasn't
that
good
in
recognizing
it.
But
let's
say:
let's
give
another
try
and
let's
try.
Let's
see,
what's
stopping
you
like
that,
so.
A
Evaluation
now
it's
eight
okay.
We
have
infused
artificial
intelligence
in
our
application.
That's
that's
the
most.
That's
almost
all
we
are
over
there
and
all
these
resources
will
be
will
be
available
for
you
to
the
logo,
so
this
light
will
be
available
to
download.
So
you
can
get
also
altar
Association.
As
you
can
see,
there
are
a
lot
of
good
point
to
start.
One
thing
that
I
want
to
point
you
out
is
this
one:
this
is
three
hour
and
a
half
courses
on
theory
about
deeply.
A
You
must
see
this
course
and
also,
let's
see,
let's
see
an
effect,
sometimes
for
the
people
that
help
me
and
get
me
inspired
for
this
course
and
to
teach
myself
artificial
intelligence,
because
this
is
something
really
really
hard
at
the
beginning.
I
hope
this
will
help
you,
and
so,
if
we
have
any
question
this
is
the
moment
we
can
get
that.