Lecture 2: Introduction to Machine Learning

Lecture 2: Introduction to Machine Learning

Machine Learning Definition

• Field of study that gives computer the ability to learn without being explicitly programmed (Arthur Samuel, 1956)

• Study of algorithms that improve their performance P at some task T with experience E (Tom Mitchell, 1998)

Well defined learning task: <P, T, E>

T: Play checkersP: % of games wonE: Playing against self

Well Defined Learning Task• Handwriting Recognition

– Task T: recognizing and classifying handwritten words within images

– Performance P: percent of words correctly classified– Training experience E: a database of written words with given

classification

Question• Suppose your email program watches which email you do and do

not mark as spam and based on that learn how to better filter spam. What is the task in this setting– Classifying emails as spam or not spam– The number of emails correctly classifying as spam/not spam– Labelling emails as spam/ not spam– Non of above: This is not a machine learning problem

Machine Learning Algorithms

• Supervised Learning Algorithms• Unsupervised Learning Algorithms

Supervised Learning

• Right answers are given for inputs• Regression refers to predicting continuous

valued output (e.g. price)

Supervised Learning

• Classification refers to predict discrete valued output (e.g. 0 or 1)

Supervised Learning

• More sophisticated features are:– Uniformity of cell size– Uniformity of cell shape, etc

QuestionSuppose you are running a company and want to develop a learning algorithm to address each of two problems:

• Problem 1: you have large inventory of identical items. You want to predict how many of items will sell over next 3 months.

• Problem 2: you would like your program to examine individual customer accounts and for each account decide if it has been hacked or not.

• Should you treat these as classification or regression problem ?– Treat both as classification problem – Treat problem 1 as classification and 2 as regression problem– Treat both as regression problem– Treat 1 as regression and 2 as classification problems

Unsupervised Learning

Unsupervised Learning Application

Unsupervised Learning Application

Unsupervised Learning Application

Figure: DNA microarray data of individuals

Unsupervised Learning Application

0 50 100 150 200 250 300 350 400 4500

50

100

150

200

250

Windows [#]

Aver

age

powe

r con

sum

ptio

nn[W

]

1000

1500

2000

2500

3000

Fridge

Fridge and computer

Fridge, computer and dishwasher

Window size = 2 minutes

Unsupervised Learning Application

0 50 100 150 200 250 300 350 400 4500

50

100

150

200

250

Windows [#]

Aver

age

powe

r con

sum

ptio

nn[W

]

1000

1500

2000

2500

3000 State: S1State: S2State: S3State: S4State: S5State: S6State: S7

Fridge

Fridge and computer

Fridge, computer and dishwasher

0 10 20 30 40 50 60 70 800

20

40

60

80

100

120

Windows [#]

Aver

age

powe

r co

nsum

ptio

n [W

]

State: S1State: S3State: S4

State sequence of fridgeS1 S4 S3

120 130 140 150 160 170 180 190 200 2100

20406080

100120140160

Windows [#]

Aver

age

powe

r co

nsum

ptio

n [W

]

State: S2State:S5

State sequence of fridge and computerS2 S5 S2 S5

310 320 330 340 350 360 370 380 3900

500

1000

1500

2000

2500

3000

Windows [#]

Aver

age

powe

r co

nsum

ptio

n [W

]

State: S2State: S5State: S6State: S7

State sequence of diswasherS6 S7 S6 S7

Window size = 2 minutes

Unsupervised Learning Applications

Unsupervised Learning Application: Cocktail Party Problem

Unsupervised Learning Application: Cocktail Party Algorithm

Question

• Of following examples, which one you address using unsupervised learning algorithm?– Given email labelled as spam/not spam, learn a spam

filter– Given a set of news articles on the web, group them into

set of articles about the same story– Given a database of customer data, automatically

discover market segments and group customer into different market segments

– Given a database of patients diagnosed as either having diabetes or not, learn to classify a new patients as either having a diabetes or not.

Ungraded Assignment

• Install Octave – an open source software or• Practice with:

– Elementary operation: add, subtract, multiplication, power, divide, etc

– Conditional operation: equal, not equal, greater, greater and equal to, etc

– Logical operations: AND, OR, XOR, etc – Variable assignment– Vectors and matrices: defining vectors and matrices,

ones, zeros, rand, eye– doc and help comand