You are on page 1of 20

What Is Machine Learning?

In machine learning we want


to teach a machine to learn
?

DATA MACHINE MODEL PREDICTION


Learning a Predictive Model
Based on Labeled Data

x = data/features for a subject


Learning a Predictive Model
Based on Labeled Data

3 12 32 50 7 2 53 32 21 65 43 39 27 6 5 19 12

x = data/features for a subject y = associated label 0/1


Learning a Predictive Model
Based on Labeled Data

3 12 32 50 7 2 53 32 21 65 43 39 27 6 5 19 12

x = data/features for a subject y = associated label 0/1

pixelated image benign malignant


Training Set

x1 y1

x2 y2

x3 y3

x4 y4

xN – 1 yN – 1

xN yN
Training Set
x = data
x1 y1

x2 y2

x3 y3

x4 y4

xN – 1 yN – 1

xN yN
Training Set
x = data y = outcome
x1 y1

x2 y2

x3 y3

x4 y4

xN – 1 yN – 1

xN yN
Training Set
x = data y = outcome
x1 y1

x2 y2
?
x3 y3

x4 y4
PREDICTION

xN – 1 yN – 1

xN yN
Will it Rain?

features outcome
Will it Rain?
xi = features
for day i features outcome
Cloud Cover Humidity Temperature Air Pressure

WEATHER
Will it Rain?
xi = features
for day i features outcome
Cloud Cover Humidity Temperature Air Pressure

0.5 80% 75 1.2

WEATHER
Will it Rain?
xi = features
for day i features outcome
Cloud Cover Humidity Temperature Air Pressure Did it Rain

0.5 80% 75 1.2 1

WEATHER RAIN
Will it Rain?
xi = features yi = 1, rains
for day i features outcome
Cloud Cover Humidity Temperature Air Pressure Did it Rain

0.5 80% 75 1.2 1

WEATHER RAIN
Will it Rain?
xi = features yi = 1, rains
for day i features outcome yi = 0, no rain
Cloud Cover Humidity Temperature Air Pressure Did it Rain

0.5 80% 75 1.2 1

WEATHER RAIN
Learned Model Parameters

Training Set
x1 y1
x2 y2
x3 y3
x4 y4

xN – 1 yN – 1
xN yN
Learned Model Parameters

Training Set Mathematical Model


x1 y1
x2 y2
x3 y3
?
x4 y4

xi1 xiM
xN – 1 yN – 1
xN yN
Learned Model Parameters

Training Set Mathematical Model


x1 y1
x2 y2
x3 y3
?
x4 y4

xi1 xiM
xN – 1 yN – 1
xN yN Learned Parameters
Learning a Predictive Model
Based on Labeled Data

3 12 32 50 7 2 53 32 21 65 43 39 27 6 5 19 12

x = data/features for a subject y = associated label 0/1

You might also like