Embed
Email

Machine Learning

Document Sample
Machine Learning
Shared by: HC11121420243
Categories
Tags
Stats
views:
0
posted:
12/14/2011
language:
pages:
12
Machine Learning



A Quick look

Sources:

• Artificial Intelligence – Russell & Norvig

• Artifical Intelligence - Luger







By: Héctor Muñoz-Avila

What Is Machine Learning?

“Logic is not the end of wisdom, it is just the beginning” --- Spock





same time

Environment Environment









System System





Action1 Action2









Knowledge Knowledge

changed

Classification

(According to the language representation)

• Symbolic

 Version Space

 Decision Trees

 Explanation-Based Learning

 …





• Sub-symbolic

 Connectionist

 Evolutionary

Version Space

Idea: Learn a concept from a group of instances, some

positive and some negative

Example: Two extremes (temptative) solutions:

•target: obj(Size,Color,Shape) too general

Size = {large, small} obj(X,Y,Z)

Color = {red, white, blue}



Shape = {ball, brick, cube}

concept space obj(X,Y,ball)

•Instances:

+:

obj(large,white,ball)

obj(small,blue,ball) obj(large,Y,ball) obj(small,Y,ball)



−:

obj(small,red,brick) obj(large,white,ball) obj(small,blue,ball) …

obj(large,blue,cube)

too specific

How Version Space Works



If we consider only positives If we consider positive and negatives



+ + +

+ +

+ +

+

+ + + +



− + − − + −

+ +









What is the role of the negative instances?

to help prevent over-generalizations

Explanation-Based learning

A C

C B

B A

C B

A B A C

?

B ? C ? B A

C A B C B

A B A C

A C



A A

B C B C

C B

A A

B C

Can we avoid making this error again?

Explanation-Based learning (2)

C

A B

?

B ? C ?

A

C A B C B

A B

A C



Possible rule: If the initial state is this and the final state is this,

don’t do that





More sensible rule: don’t stack anything above a block, if the block has

to be free in the final state

Evolutionary Approaches

Idea: Biological analogy on how populations of species evolve over

generations



Step 1: start with a population (each member is a candidate solution)





Step 2: Create the next generation by considering evolutionary

operations on the population from the previous generation (e.g.,

mutation) and a fitness function (only the more fit get to contribute to

the next generation)







Continue the process until a certain condition is reached

The Genetic Algorithm

t0

Initialize the population P(t)

Crossover

While the termination condition is not met do Mutation

{ Inversion

evaluate fitness of each member of P(t) exchange

select members of P(t) based on fitness

produce the offspring of pairs of selected members using genetic

operators

replace, based on fitness, candidates of P(t) based on this offspring

tt+1

}

Non-selected members are not necessarily eliminated

Example: CNF-satisfaction

A conjunctive normal form (CNF) is a Boolean expression

consisting of one or more disjunctive formulas connected by an

AND symbol (). A disjunctive formula is a collection of one or

more (positive and negative) literals connected by an OR

symbol ().



Example:

(a)  (¬ a  ¬b  c  d)  (¬c  ¬d)  (¬d)



Problem (CNF-satisfaction): Give an algorithm that

receives as input a CNF form and returns Boolean

assignments for each literal in form such that form is true

Example (above):

a  true, b  false, c  true, d  false

CNF as a Genetic Algorithm

• A potential solution is a true/false assignment to the 4 variables a,

b, c, and d in the formula: 1010 means that a and c are true and b

and d are false



• In particular, a solution for (a)  (¬ a  ¬b  c  d)  (¬c  ¬d)  (¬d) is

1001





• Nice: all 4 genetic operations applied on any potential solutions

will result in a potential solutions (in other problems or other

representations of this problem this may not be the case)



•Fitness: for 0101 and 1001: which is a more suitable solution? 1001

1 2

•Fitness value? # of disjunctions in the formula that are made true

The Genetic Algorithm for CNF

t0 N randomly generated strings of 4 integers

Initialize the population P(t) Solution has not been found

While the termination condition is not met do

{ # of disjunctions in the formula

evaluate fitness of each member of P(t) that are made true

select members of P(t) based on fitness Select top 30%

produce the offspring of pairs of selected members using genetic

operators Select among the 4 operations randomly

replace, based on fitness, candidates of P(t) based on this offspring

tt+1

Top N candidates

}


Related docs
Other docs by HC11121420243
BAP Minutes- October 3, 2007
Views: 0  |  Downloads: 0
Arithmetic Mean
Views: 0  |  Downloads: 0
2010
Views: 0  |  Downloads: 0
EDN 523: Research Designs
Views: 0  |  Downloads: 0
PSYC230: Social Psychology
Views: 6  |  Downloads: 0
MACHINERY SALES COMPANY, INC
Views: 0  |  Downloads: 0
By registering with docstoc.com you agree to our
privacy policy

You are almost ready to download!

You are almost ready to download!