Machine Learning: Lecture 2

Concept Learning
and
Version Spaces

thanks to Brian Pardo (http://bryanpardo.com) for the illustrations

on slides 17, 22, 25 and 26
1
What is a Concept? (1)
 A Concept is a a subset of objects or events defined
over a larger set [Example: The concept of a bird is the subset
of all objects (i.e., the set of all things or all animals) that
belong to the category of bird.]

Things Birds
Alternatively, a concept is a boolean-valued function
Animals
defined over this larger set [Example: a function defined
Cars
over all animals whose value is true for birds and false for
every other animal].

2
What is a Concept? (2)
Given X the set of all examples.
A concept C is a subset of X.
A training example T is a subset of X such
that some examples of T are elements of C
(the positive examples) and some examples
are not elements of C (the negative
examples)

3
What is Concept-Learning? (1)
Given a set of examples labeled as members or
non-members of a concept, concept-learning
consists of automatically inferring the general
definition of this concept.

In other words, concept-learning consists of

approximating a boolean-valued function from
training examples of its input and output.

4
What is Concept Learning? (2)
 Learning:
Learning
{<xi,yi>}   f: X Y
system
with i=1..n,
• xi T, yi  Y (={0,1})
• yi= 1, if x1 is positive ( C)
• yi= 0, if xi is negative ( C)
 Goals of learning:
f must be such that for all xj  X (not only  T) -
f(xj) =1 if xj  C
- f(xj) = 0, if xj  C

5
 The Problem of Induction:
Computer Science’s Answer
 Problem: As previously noted by philosophers, the task of
induction is not well formulated. In computer science the problem
can be thought of as follows: there exists an infinite number of
functions that satisfy the goal  It is necessary to find a way to
constrain the search space of f.
 Definitions:

The set of all fs that satisfy the goal is called hypothesis space.

The constraints on the hypothesis space is called the inductive
bias.

There are two types of inductive bias:
• The hypothesis space restriction bias
• The preference bias

6
 Inductive Biases (1)
 Hypothesis space restriction bias  We restrain the
language of the hypothesis space. Examples:
 k-DNF: We restrict f to the set of Disjunctive Normal form
formulas having an arbitrary number of disjunctions but at
most, k conjunctive in each conjunctions.
 K-CNF: We restrict f to the set of Conjunctive Normal Form
formulas having an arbitrary number of conjunctions but with
at most, k disjunctive in each disjunction.
 Properties of that type of bias:

Positive: Learning will by simplified (Computationally)

Negative: The language can exclude the “good” hypothesis.

7
Inductive Biases (2)
 Preference Bias: It is an order or unit of measure
that serves as a base to a relation of preference in
the hypothesis space.
 Examples:
 Occam’s razor: We prefer a simple formula for f.
 Principle of minimal description length (An
extension of Occam’s Razor): The best hypothesis
is the one that minimise the total length of the
hypothesis and the description of the exceptions
to this hypothesis.
8
Using Biases for Learning (1)
 How to implement learning with these bias?

 Hypothesis space restriction bias:


Given:
• A set S of training examples
• A set of restricted hypothesis, H

Find: An hypothesis f  H that minimizes the
number of incorrectly classified training
examples of S.
9
 Using Biases for Learning (2)
 Preference Bias:

Given:
• A set S of training examples
• An order of preference better(f1, f2) for all the hypothesis
space (H) functions.

Find: the best hypothesis f  H (using the “better” relation)
that minimises the number of training examples S incorrectly
classified.
 Search techniques:

Heuristic search

Hill Climbing

Simulated Annealing et Genetic Algorithm

10
Example of a Concept Learning task
 Concept: Good Days for Water Sports (values: Yes,
No)
 Attributes/Features:

Sky (values: Sunny, Cloudy, Rainy)

AirTemp (values: Warm, Cold)

Humidity (values: Normal, High)

Wind (values: Strong, Weak)

Water (Warm, Cool)

Forecast (values: Same, Change)
 Example of a Training Point:
<Sunny, Warm, High, Strong, Warm, Same, Yes>
class

11
 Example of a Concept Learning task
Database:
Day Sky AirTemp Humidity Wind Water Forecast WaterSport
1 Sunny Warm Normal Strong Warm Same Yes
2 Sunny Warm High Strong Warm Same Yes
3 Rainy Cold High Strong Warm Change No
4 Sunny Warm High Strong Cool Change Yes

Chosen Hypothesis Representation: class

Conjunction of constraints on each attribute where:
• “?” means “any value is acceptable”
• “0” means “no value is acceptable”
Example of a hypothesis: <?,Cold,High,?,?,?>
(If the air temperature is cold and the humidity high then
it is a good day for water sports)
12
Example of a Concept Learning task
Goal: To infer the “best” concept-description from the
set of all possible hypotheses (“best” means “which best
generalizes to all (known or unknown) elements of the
instance space”. . concept-learning is an
ill-defined task)
Most General Hypothesis: Everyday is a good day for
water sports <?,?,?,?,?,?>
Most Specific Hypothesis: No day is a good day for
water sports <0,0,0,0,0,0>

13
Terminology and Notation
 The set of items over which the concept is defined is called the set of instances
(denoted by X)
 The concept to be learned is called the Target Concept (denoted by c: X--> {0,1})
 The set of Training Examples is a set of instances, x, along with their target
concept value c(x).
 Members of the concept (instances for which c(x)=1) are called positive examples.
 Nonmembers of the concept (instances for which c(x)=0) are called negative
examples.
 H represents the set of all possible hypotheses. H is determined by the human
designer’s choice of a hypothesis representation.
 The goal of concept-learning is to find a hypothesis h:X --> {0,1} such that
h(x)=c(x) for all x in X.

14
Concept Learning as Search
 Concept Learning can be viewed as the task of
searching through a large space of hypotheses
implicitly defined by the hypothesis
representation.
 Selecting a Hypothesis Representation is an
important step since it restricts (or biases) the
space that can be searched. [For example, the
hypothesis “If the air temperature is cold or the humidity
high then it is a good day for water sports” cannot be
expressed in our chosen representation.]

15
 General to Specific Ordering of
Hypotheses I
 Definition: Let hj and hk be boolean-valued functions
defined over X. Then hj is more-general-than-or-equal-
to hk iff For all x in X, [(hk(x) = 1) --> (hj(x)=1)]
 Example:

h1 = <Sunny,?,?,Strong,?,?>

h2 = <Sunny,?,?,?,?,?>
Every instance that are classified as positive by h1 will
also be classified as positive by h2 in our example data
set. Therefore h2 is more general than h1.

16
General to Specific Ordering of
Hypotheses II

from Bryan Pardo, EECS 349, Machine Learning, Fall 2009

17
Find-S, a Maximally Specific
Hypothesis Learning Algorithm
 Initialize h to the most specific hypothesis in H
For each positive training instance x

For each attribute constraint ai in h
If the constraint ai is satisfied by x
then do nothing
else replace ai in h by the next more general
constraint that is satisfied by x
Output hypothesis h

18
Shortcomings of Find-S
 Although Find-S finds a hypothesis consistent
with the training data, it does not indicate
whether that is the only one available
 Is it a good strategy to prefer the most specific
hypothesis?
 What if the training set is inconsistent (noisy)?
 What if there are several maximally specific
consistent hypotheses? Find-S cannot backtrack!

19
 Version Spaces and the
Candidate-Elimination Algorithm
 Definition: A hypothesis h is consistent with a set of
training examples D iff h(x) = c(x) for each example
<x,c(x)> in D.
 Definition: The version space, denoted VS_H,D, with
respect to hypothesis space H and training examples D,
is the subset of hypotheses from H consistent with the
training examples in D.
 NB: While a Version Space can be exhaustively
enumerated, a more compact representation is preferred.

20
 A Compact Representation for
Version Spaces
 Instead of enumerating all the hypotheses consistent with a
training set, we can represent its most specific and most general
boundaries. The hypotheses included in-between these two
boundaries can be generated as needed.
 Definition: The general boundary G, with respect to
hypothesis space H and training data D, is the set of maximally
general members of H consistent with D.
 Definition: The specific boundary S, with respect to hypothesis
space H and training data D, is the set of minimally general (i.e.,
maximally specific) members of H consistent with D.

21
A Compact Representation for
Version Spaces: An example

22
Version Spaces: Definitions
 Given C1 and C2, two concepts represented by sets of
examples. If C1  C2, then C1 is a specialisation of C2 and
C2 is a generalisation of C1.
 C1 is also considered more specific than C2
 Example: The set off all blue triangles is more specific than
the set of all the triangles.
 C1 is an immediate specialisation of C2 if there is no concept
that are a specialisation of C2 and a generalisation of C1.
 A version space define a graph where the nodes are concepts
and the arcs specify that a concept is an immediate
specialisation of another one.

23
Candidate-Elimination Learning
Algorithm

 The candidate-Elimination algorithm

computes the version space containing all
(and only those) hypotheses from H that are
consistent with an observed sequence of
training examples.

24
Version Space Example

25
Version Space Example (cont’d)

26
Remarks on Version Spaces and
Candidate-Elimination
 The version space learned by the Candidate-Elimination
Algorithm will converge toward the hypothesis that correctly
describes the target concept provided: (1) There are no errors in
the training examples; (2) There is some hypothesis in H that
correctly describes the target concept.
 Convergence can be speeded up by presenting the data in a
strategic order. The best examples are those that satisfy exactly
half of the hypotheses in the current version space.
 Version-Spaces can be used to assign certainty scores to the
classification of new examples

27
Inductive Bias I: A Biased
Hypothesis Space
Database:
Day Sky AirTemp Humidity Wind Water Forecast WaterSport
1 Sunny Warm Normal Strong Cool Change Yes
2 Cloudy Warm Normal Strong Cool Change Yes
3 Rainy Warm Normal Strong Cool Change No
class
Given our previous choice of the hypothesis space
representation, no hypothesis is consistent with the
above database: we have BIASED the learner to
consider only conjunctive hypotheses
28
Inductive Bias II: An Unbiased
Learner
 In order to solve the problem caused by the bias of the
hypothesis space, we can remove this bias and allow the
hypotheses to represent every possible subset of instances.
The previous database could then be expressed as:
<Sunny, ?,?,?,?,?> v <Cloudy,?,?,?,?,?,?>
 However, such an unbiased learner is not able to
generalize beyond the observed examples!!!! All the non-
observed examples will be well-classified by half the
hypotheses of the version space and misclassified by the
other half.

29
Inductive Bias III: The Futility of
Bias-Free Learning
 Fundamental Property of Inductive Learning A
learner that makes no a priori assumptions regarding the
identity of the target concept has no rational basis for
classifying any unseen instances.
 We constantly have recourse to inductive biases
Example: we all know that the sun will rise tomorrow.
Although we cannot deduce that it will do so based on
the fact that it rose today, yesterday, the day before, etc.
(see the philosophical basis of induction) , we do take
this leap of faith or use this inductive bias, naturally!

30
Inductive Bias IV: A Definition
 Consider a concept-learning algorithm L for the set of
instances X. Let c be an arbitrary concept defined over
X, and let Dc = {<x,c(x)>} be an arbitrary set of
training examples of c. Let L(xi,Dc) denote the
classification assigned to the instance xi by L after
training on the data Dc. The inductive bias of L is any
minimal set of assertions B such that for any target
concept c and corresponding training examples Dc
(For all xi in X) [(B ^Dc^xi) |-- L(xi,Dc)]

31
Ranking Inductive Learners
according to their Biases
Weak

Rote-Learner: This system simply memorizes
the training data and their classification--- No
generalization is involved.
Bias
Strength

Candidate-Elimination: New instances are
classified only if all the hypotheses in the
version space agree on the classification

Find-S: New instances are classified using the
most specific hypothesis consistent with the
training data
Strong
32