Neural
Networks:
Representa1on
Non-‐linear
hypotheses
Machine
Learning
Non-‐linear
Classifica/on
x2
x1
size
#
bedrooms
#
floors
age
Andrew
Ng
What
is
this?
You
see
this:
But
the
camera
sees
this:
Andrew
Ng
Computer
Vision:
Car
detec/on
Cars
Not
a
car
Tes1ng:
What
is
this?
Andrew
Ng
pixel
1
Learning
Algorithm
pixel
2
Raw
image
pixel
2
Cars
pixel
1
“Non”-‐Cars
Andrew
Ng
pixel
1
Learning
Algorithm
pixel
2
Raw
image
pixel
2
Cars
pixel
1
“Non”-‐Cars
Andrew
Ng
pixel
1
Learning
Algorithm
pixel
2
Raw
image
50
x
50
pixel
images→
2500
pixels
pixel
2
(7500
if
RGB)
pixel
1
intensity
pixel
2
intensity
pixel
2500
intensity
Cars
pixel
1
Quadra1c
features
(
):
≈3
million
“Non”-‐Cars
features
Andrew
Ng
Neural
Networks:
Representa1on
Neurons
and
the
brain
Machine
Learning
Neural
Networks
Origins:
Algorithms
that
try
to
mimic
the
brain.
Was
very
widely
used
in
80s
and
early
90s;
popularity
diminished
in
late
90s.
Recent
resurgence:
State-‐of-‐the-‐art
technique
for
many
applica1ons
Andrew
Ng
The
“one
learning
algorithm”
hypothesis
Auditory
Cortex
Auditory
cortex
learns
to
see
[Roe
et
al.,
1992]
Andrew
Ng
The
“one
learning
algorithm”
hypothesis
Somatosensory
Cortex
Somatosensory
cortex
learns
to
see
[Me1n
&
Frost,
1989]
Andrew
Ng
Sensor
representa/ons
in
the
brain
Seeing
with
your
tongue
Human
echoloca1on
(sonar)
Hap1c
belt:
Direc1on
sense
Implan1ng
a
3rd
eye
[BrainPort;
Welsh
&
Blasch,
1997;
Nagel
et
al.,
2005;
Constan1ne-‐Paton
&
Law,
2009]
Andrew
Ng
Neural
Networks:
Representa1on
Model
representa1on
I
Machine
Learning
Neuron
in
the
brain
Andrew
Ng
Neurons
in
the
brain
[Credit:
US
Na1onal
Ins1tutes
of
Health,
Na1onal
Ins1tute
on
Aging]
Andrew
Ng
Neuron
model:
Logis/c
unit
Sigmoid
(logis1c)
ac1va1on
func1on.
Andrew
Ng
Neural
Network
Layer
1
Layer
2
Layer
3
Andrew
Ng
Neural
Network
“ac1va1on”
of
unit
in
layer
matrix
of
weights
controlling
func1on
mapping
from
layer
to
layer
If
network
has
units
in
layer
,
units
in
layer
,
then
will
be
of
dimension
.
Andrew
Ng
Neural
Networks:
Representa1on
Model
representa1on
II
Machine
Learning
Forward
propaga/on:
Vectorized
implementa/on
Add
.
Andrew
Ng
Neural
Network
learning
its
own
features
Layer
1
Layer
2
Layer
3
Andrew
Ng
Other
network
architectures
Layer
1
Layer
2
Layer
3
Layer
4
Andrew
Ng
Neural
Networks:
Representa1on
Examples
and
intui1ons
I
Machine
Learning
Non-‐linear
classifica/on
example:
XOR/XNOR
,
are
binary
(0
or
1).
x2
x2
x1
x1
Andrew
Ng
Simple
example:
AND
1.0
0
0
0
1
1
0
1
1
Andrew
Ng
Example:
OR
func/on
-‐10
20
0
0
20
0
1
1
0
1
1
Andrew
Ng
Neural
Networks:
Representa1on
Examples
and
intui1ons
II
Machine
Learning
Nega/on:
0
1
Andrew
Ng
PuPng
it
together:
-‐30
10
-‐10
20
-‐20
20
20
-‐20
20
0
0
0
1
1
0
1
1
Andrew
Ng
Neural
Network
intui/on
Layer
1
Layer
2
Layer
3
Layer
4
Andrew
Ng
HandwriRen
digit
classifica/on
[Courtesy
of
Yann
LeCun]
Andrew
Ng
Neural
Networks:
Representa1on
Mul1-‐class
classifica1on
Machine
Learning
Andrew
Ng
Mul/ple
output
units:
One-‐vs-‐all.
Pedestrian
Car
Motorcycle
Truck
Want
,
,
,
etc.
when
pedestrian
when
car
when
motorcycle
Andrew
Ng
Mul/ple
output
units:
One-‐vs-‐all.
Want
,
,
,
etc.
when
pedestrian
when
car
when
motorcycle
Training
set:
one
of
,
,
,
pedestrian
car
motorcycle
truck
Andrew
Ng