8/12/2019 19-mlintro

1/29

Machine Learning

Mausam

(based on slides by Tom Mitchell, OrenEtzioni and Pedro Domingos)

8/12/2019 19-mlintro

2/29

What Is Machine Learning?

A computer program is said to learn fromexperience E with respect to some class oftasks T and a performance measure P if itimproves performance on T (according to P)with more E.

2

8/12/2019 19-mlintro

3/29

Traditional Programming

Machine Learning

3

Computer Data

ProgramOutput

Computer DataOutput

Program

8/12/2019 19-mlintro

4/29

Why Bother with Machine Learning?

Btw, Machine Learning ~ Data Mining

Necessary for AI Learn concepts that people dont have time

for (drowning in datastarved forknowledge)

Mass customization (adapt software to each) Super-human learning/discovery

4

8/12/2019 19-mlintro

5/29

Quotes A break through in machine learning would be worth ten Microsofts

(Bill Gates) Machine learning is the next Internet

(Tony Tether, Former Director, DARPA) Machine learning is the hot new thing

(John Hennessy, President, Stanford) Web rankings today are mostly a matter of machine learning

(Prabhakar Raghavan, Dir. Research, Yahoo)

Machine learning is going to result in a real revolution(Greg Papadopoulos, CTO, Sun)

5

8/12/2019 19-mlintro

6/29

Inductive Learning

Given examples of a function (X, F(X)) Predict function F(X) for new examples X

Discrete F(X): Classification Continuous F(X): Regression F(X)= Probability( X ): Probability estimation

6

8/12/2019 19-mlintro

7/29

Training Data Versus Test

Terms: data, examples, and instances usedinterchangeably

Training data: data where the labels are given Test data: data where the labels are known

but not givenWhich do you use to measure performance?Cross validation

7

8/12/2019 19-mlintro

8/29

Basic Setup Input:

Labeled training examples Hypothesis space H

Output: hypothesis h in H that is consistentwith the training data & (hopefully) correctly

classifies test data.

8

8/12/2019 19-mlintro

9/29

The new Machine Learning

Old New

Small data sets (100s of examples) Massive (10^6 to 10^10)

Hand-labeled data Automatically labeled; semi supervised;labeled by crowds

Hand-coded algorithms WEKA package downloaded over1,000,000 times

9

8/12/2019 19-mlintro

10/29

ML in a Nutshell

10^5 machine learning algorithms Hundreds new every year

Every algorithm has three components: Hypothesis space possible outputs Search strategy---strategy for exploring space Evaluation

10

8/12/2019 19-mlintro

11/29

Hypothesis Space (Representation)

Decision trees Sets of rules / Logic programs Instances Graphical models (Bayes/Markov nets) Neural networks Support vector machines Model ensembles Etc.

11

8/12/2019 19-mlintro

12/29

Metrics for Evaluation Accuracy Precision and recall Squared error

Likelihood Posterior probability Cost / Utility Margin

Etc.

Based on Data

12

8/12/2019 19-mlintro

13/29

Search Strategy

Greedy (depth-first, best-first, hill climbing)

Exhaustive

Optimize an objective function

More

13

8/12/2019 19-mlintro

14/29

Types of Learning

Supervised (inductive) learning Training data includes desired outputs

Unsupervised learning Training data does not include desired outputs

Semi-supervised learning Training data includes a few desired outputs

Reinforcement learning Rewards from sequence of actions

14

8/12/2019 19-mlintro

15/29

15

8/12/2019 19-mlintro

16/29

16

8/12/2019 19-mlintro

17/29

17

Why Learning? Learning is essential for unknown environments

e.g., when designer lacks omniscience

Learning is necessary in dynamic environments Agent can adapt to changes in environment not foreseen at

design time

Learning is useful as a system construction method Expose the agent to reality rather than trying to approximate

it through equations etc.

Learning modifies the agent's decision mechanisms toimprove performance

8/12/2019 19-mlintro

18/29

18

8/12/2019 19-mlintro

19/29

19

8/12/2019 19-mlintro

20/29

20

8/12/2019 19-mlintro

21/29

8/12/2019 19-mlintro

22/29

8/12/2019 19-mlintro

23/29

Inductive Bias

Need to make assumptions Experience alone doesnt allow us to make

conclusions about unseen data instances

Two types of bias: Restriction: Limit the hypothesis space

(e.g., nave Bayes) Preference: Impose ordering on hypothesis space

(e.g., decision tree)

23

8/12/2019 19-mlintro

24/29

24

Inductive learning example Construct h to agree with f on training set

h is consistent if it agrees with f on all trainingexamples

E.g., curve fitting (regression):

x = Input data point

(training example)

8/12/2019 19-mlintro

25/29

25

Inductive learning example

h = Straight line?

8/12/2019 19-mlintro

26/29

26

Inductive learning example What about a quadratic function?

What aboutthis littlefella?

8/12/2019 19-mlintro

27/29

27

Inductive learning exampleFinally, a function that satisfies all!

8/12/2019 19-mlintro

28/29

28

But so does this one

Inductive learning example

8/12/2019 19-mlintro

29/29

29

Ockhams Razor Principle

Ockhams razor: prefer the simplest hypothesis consistent with data

Related to KISS principle (keep it simple stupid)Smooth blue function preferable over wiggly yellow oneIf noise known to exist in this data, even linear might be

better (the lowest x might be due to noise)