UNIT I: Concept Learning

• Learning from examples

• General-to-specific ordering over hypotheses
• Version Spaces and candidate elimination
algorithm
• Picking new examples
• The need for inductive bias

1
Features from Computer View

Eyes Nose Head Fcolor Hair? Smile?

Round Triangle Round Purple Yes Yes

Square Square Square Green Yes No

Square Triangle Round Yellow Yes Yes

Round Triangle Round Green No No

Square Square Round Yellow Yes Yes

2
Representing Hypotheses
Many possible representations for hypotheses h
Idea: h as conjunctions of constraints on features
Each constraint can be:
– a specific value (e.g., Nose = Square)
– don’t care (e.g., Eyes = ?)
– no value allowed (e.g., Water=Ø)
For example,
Eyes Nose Head Fcolor Hair?
<Round, ?, Round, ?, No>
?

3
Prototypical Concept Learning Task
Given:
– Instances X: Faces, each described by the attributes
Eyes, Nose, Head, Fcolor, and Hair?
– Target function c: Smile? : X -> { no, yes }
– Hypotheses H: Conjunctions of literals such as
<?,Square,Square,Yellow,?>
– Training examples D: Positive and negative examples
of the target function
 x1 , c( x1 ) ,  x2 , c( x2 ) ,...,  xm , c( xm ) 
Determine: a hypothesis h in H such that h(x)=c(x)
for all x in D.

4
Inductive Learning Hypothesis
Any hypothesis found to approximate the target
function well over a sufficiently large set of
training examples will also approximate the target
function well over other unobserved examples.

• What are the implications?

• Is this reasonable?
• What (if any) are our alternatives?
• What about concept drift (what if our views/tastes
change over time)?

5
Instances, Hypotheses, and More-General-Than

Instances X Hypotheses H

General

h3

h1 h2
Specific

x 1 =<Round,Square,Square,Purple,Yes> h 1 =<Round,?,Square,?,?>
x 2 =<Round,Square,Round,Green,Yes> h 2 =<Round,?,?,?,Yes>
h 3 =<Round,?,?,?,?>

6
Find-S Algorithm
1. Initialize h to the most specific hypothesis in H
2. For each positive training instance x
For each attribute constraint ai in h
IF the constraint ai in h is satisfied by x THEN
do nothing
ELSE
replace ai in h by next more general constraint satisfied by x

3. Output hypothesis h

7
Hypothesis Space Search by Find-S
Instances X Hypotheses H
h5
x
General
2
x x h 3,4

1 3

x
5

x h 1,2
4 Specific
h0
h 0 =<  >
x 1 =<Round,Triangle,Round,Purple,Yes> + h 1 =<Round,Triangle,Round,Purple,Yes>
x 2 =<Square,Square,Square,Green,Yes> - h 2 =<Round,Triangle,Round,Purple,Yes>
x 3 =<Square,Triangle,Round,Yellow,Yes> + h 3 =<?,Triangle,Round,?,Yes>
x 4 =<Round,Triangle,Round,Green,No> - h 4 =<?,Triangle,Round,?,Yes>
x 5 =<Square,Square,Round,Yellow,Yes> + h 5 =<?,?,Round,?,Yes>

8
Complaints about Find-S
• Cannot tell whether it has learned concept
• Cannot tell when training data inconsistent
• Picks a maximally specific h (why?)
• Depending on H, there might be several!

• How do we fix this?

9
The List-Then-Eliminate Algorithm
1. Set VersionSpace equal to a list containing every
hypothesis in H
2. For each training example, <x,c(x)>
remove from VersionSpace any hypothesis h for which
h(x) != c(x)
3. Output the list of hypotheses in VersionSpace

• But is listing all hypotheses reasonable?

• How many different hypotheses in our simple
problem?
– How many not involving “?” terms?
10
Version Spaces
A hypothesis h is consistent with a set of training
examples D of target concept c if and only if
h(x)=c(x) for each training example in D.
Consistent(h, D)  (  x, c( x)   D) h( x)  c( x)
The version space, VSH,D, with respect to hypothesis
space H and training examples D, is the subset of
hypotheses from H consistent with all training
examples in D.
VS H , D  {h  H | Consistent(h, D)}

11
Example Version Space

G: { <?,?,Round,?,?> <?,Triangle,?,?,?> }

<?,?,Round,?,Yes> <?,Triangle,Round,?,?> <?,Triangle,?,?,Yes>

S: { <?,Triangle,Round,?,Yes> }

12
Representing Version Spaces
The General boundary, G, of version space VSH,D is
the set of its maximally general members.

The Specific boundary, S, of version space VSH,D is

the set of its maximally specific members.

Every member of the version space lies between

these boundaries
VS H , D  {h  H | (s  S )(g  G )( g  h  s )}
where x  y means x is more general or equal to y
13
Candidate Elimination Algorithm
G = maximally general hypotheses in H
S = maximally specific hypotheses in H

For each training example d, do

If d is a positive example
Remove from G any hypothesis that does not include d
For each hypothesis s in S that does not include d
Remove s from S
Add to S all minimal generalizations h of s such that
1. h includes d, and
2. Some member of G is more general than h
Remove from S any hypothesis that is more general
than another hypothesis in S

14
Candidate Elimination Algorithm (cont)
For each training example d, do (cont)
If d is a negative example
Remove from S any hypothesis that does include d
For each hypothesis g in G that does include d
Remove g from G
Add to G all minimal generalizations h of g such that
1. h does not include d, and
2. Some member of S is more specific than h
Remove from G any hypothesis that is less general
than another hypothesis in G

If G or S ever becomes empty, data not consistent (with H)

15
Example Trace
G1 G0: { <?,?,?,?,?> }
X2=<S,S,S,G,Y> -

G2: { <R,?,?,?,?>, <?,T,?,?,?>, <?,?,R,?,?>, <?,?,?,P,?> } G3

G5 G4: { <?,T,?,?,Y>, <?,?,R,?,Y> } X4=<R,T,R,G,N> -

S5: { <?,?,R,?,Y> } X5=<S,S,R,Y,Y> +

S4 S3: { <?,T,R,?,Y> } X3=<S,T,R,Y,Y> +

S2 S1: { <R,T,R,P,Y> } X1=<R,T,R,P,Y> +

S0: { <Ø,Ø,Ø,Ø,Ø> }
16
What Training Example Next?

G: { <?,?,Round,?,?> <?,Triangle,?,?,?> }

<?,?,Round,?,Yes> <?,Triangle,Round,?,?> <?,Triangle,?,?,Yes>

S: { <?,Triangle,Round,?,Yes> }

17
How Should These Be Classified?
G: { <?,?,Round,?,?> <?,Triangle,?,?,?> }

<?,?,Round,?,Yes> <?,Triangle,Round,?,?> <?,Triangle,?,?,Yes>

S: { <?,Triangle,Round,?,Yes> }

? ? ?
18
What Justifies this Inductive Leap?
+ < Round, Triangle, Round, Purple, Yes >
+ < Square, Triangle, Round, Yellow, Yes >

S: < ?, Triangle, Round, ?, Yes >

Why believe we can classify the unseen?

< Square, Triangle, Round, Purple, Yes > ?

19
An UN-Biased Learner
Idea: Choose H that expresses every teachable
concept (i.e., H is the power set of X)
Consider H’ = disjunctions, conjunctions, negations
over previous H.
For example:
 ?, Triangle , Round , ?, Yes    Square, Square, ?, Purple, ? 

What are S, G, in this case?

20
Inductive Bias
Consider
– concept learning algorithm L
– instances X, target concept c
– training examples Dc={<x,c(x)>}
– let L(xi,Dc) denote the classification assigned to the
instance xi by L after training on data Dc.
Definition:
The inductive bias of L is any minimal set of assertions B
such that for any target concept c and corresponding
training examples Dc

(xi  X )[( B  Dc  xi ) L( xi , Dc )]
where A B means A logically entails B
21
Inductive Systems and Equivalent Deductive Systems

Inductive System
Candidate Classification of
Training examples new instance, or
Elimination
Algorithm "don't know"
New instance
Using Hypothesis
Space H

Equivalent Deductive System

Training examples Classification of
new instance, or
New instance "don't know"
Theorem Prover
Assertion "H
contains hypothesis"

22