Machine Learning: Lecture 9

Rule Learning /
Inductive Logic Programming /
Association Rules

1
Learning Rules
 One of the most expressive and human readable representations for
learned hypotheses is sets of production rules (if-then rules).
 Rules can be derived from other representations (e.g., decision trees) or they
can be learned directly. Here, we are concentrating on the direct method.
 An important aspect of direct rule-learning algorithms is that they can learn
sets of first-order rules which have much more representational power than
the propositional rules that can be derived from decision trees.
 Rule Learnng also allows the incorporation of background knowledge nto
the process.
 Learning rules is also useful for the data mining task of association rules
mining.

2
Propositional versus First-Order Logic
 Propositional Logic does not include variables and thus cannot
express general relations among the values of the attributes.
 Example 1: in Propositional logic, you can write: IF
(Father1=Bob) ^ (Name2=Bob)^ (Female1=True) THEN
Daughter1,2=True.
This rule applies only to a specific family!
 Example 2: In First-Order logic, you can write: IF
Father(y,x) ^ Female(y), THEN Daughter(x,y)
This rule (which you cannot write in Propositional Logic) applies to
any family!

3
Learning Propositional versus
First-Order Rules
 Both approaches to learning are useful as they address different
types of learning problems.
 Like Decision Trees, Feedforward Neural Nets and IBL systems,
Propositional Rule Learning systems are suited for problems in
which no substantial relationship between the values of the
different attributes needs to be represented.
 In First-Order Learning Problems, the hypotheses that must be
represented involve relational assertions that can be conveniently
expressed using first-order representations such as horn clauses
(H <- L1 ^…^Ln).

4
Learning Propositional Rules:
Sequential Covering Algorithms
Sequential-Covering(Target_attribute, Attributes, Examples,
Threshold)
 Learned_rules <-- { }
 Rule <-- Learn-one-rule(Target_attribute, Attributes, Examples)
 While Performance(Rule, Examples) > Threshold, do

Learned_rules <-- Learned_rules + Rule

Examples <-- Examples -{examples correctly classified by
Rule}

Rule <-- Learn-one-rule(Target_attribute, Attributes,
Examples)
 Learned_rules <-- sort Learned_rules according to
Performance over Examples
 Return Learned_rules

5
 Learning Propositional Rules:
Sequential Covering Algorithms
 The algorithm is called a sequential covering algorithm because it
sequentially learns a set of rules that together cover the whole set of
positive examples.
 It has the advantage of reducing the problem of learning a disjunctive
set of rules to a sequence of simpler problems, each requiring that a
single conjunctive rule be learned.
 The final set of rules is sorted so that the most accurate rules are
considered first at classification time.
 However, because it does not backtrack, this algorithm is not
guaranteed to find the smallest or best set of rules ---> Learn-one-rule
must be very effective!

6
Learning Propositional Rules:
Learn-one-rule
General-to-Specific Search:
 Consider the most general rule (hypothesis) which
matches every instances in the training set.
 Repeat

Add the attribute that most improves rule
performance measured over the training set.
 Until the hypothesis reaches an acceptable level of
performance.
General-to-Specific Beam Search (CN2):
 Rather than considering a single candidate at each
search step, keep track of the k best candidates.

7
Comments and Variations regarding
the Basic Rule Learning Algorithms
 Sequential versus Simultaneous covering: sequential covering
algorithms (CN2) make a larger number of independent choices
than simultaneous covering ones (ID3).
 Direction of the search: CN2 uses a general-to-specific search
strategy. Other systems (GOLEM) uses a specific to general
search strategy. General-to-specific search has the advantage of
having a single hypothesis from which to start.
 Generate-then-test versus example-driven: CN2 is a generate-
then-test method. Other methods (AQ, CIGOL) are example-
driven. Generate-then-test systems are more robust to noise.

8
Comments and Variations regarding the
Basic Rule Learning Algorithms,Cont’d
 Post-Pruning: pre-conditions can be removed
from the rule whenever this leads to improved
performance over a set of pruning examples
distinct from the training set.
 Performance measure: different evaluation can be
used. Example: relative frequency (AQ), m-
estimate of accuracy (certain versions of CN2)
and entropy (original CN2).

9
Example: RIPPER (this and the next three slides are
borrowed from E. Alpaydin, Lecture Notes for An
Introduction to machine Learning, 2004, MIT Press.
(Chapter 9)).

There are two kinds of loop in the Ripper algorithm:

Outer loop : adding one rule at a time to the rule base
Inner loop : adding one condition at a time to the
current rule
Conditions are added to the rule to maximize an
information gain measure.
Conditions are added to the rule until it covers no
negative example.

10
 O(Nlog2N)

DL: description length of

the rule base
The description length of a rule base
= (the sum of the description lengths
of all the rules in the rule base)
+ (the description of the instances
not covered by the rule base)

11
12
 Ripper Algorithm
 In Ripper, conditions are added to the rule to

Maximize an information gain measure

N' N
( R ' , R )  s  (log 2   log 2  )
Gainrule
• R : the original
N'
• R’ : the candidate rule after adding a condition N
• N (N’): the number of instances that are covered by R (R’)
• N+ (N’+): the number of true positives in R (R’)
• s : the number of true positives in R and R’ (after adding the condition)

Until it covers no negative example
p and n : the number of true and false
positives respectively.

pn
rvm( R )  1
pn
Rule value metric 13
 Incorporating Background Knowledge into the
Learning Process:
Induction as Inverted Deduction
 Let D be a set of training examples, each of the form <xi,f(xi)>. Then,
learning is the problem of discovering a hypothesis h, such that the
classification f(xi) of each training instance xi follows deductively from the
hypothesis h, the description of xi and any other background knowledge B
known to the system.
Example:
 xi: Male(Bob), Female(Sharon), Father(Sharon, Bob)
 f(xi): Child(Bob, Sharon)
 B: Parent(u,v) <-- Father(v,u)
 we want to find h s.t., (B^h^xi) |-- f(xi).
h1: Child(u,v) <-- Father(v,u) h2:
Child(u,v) <-- Parent(v,u)
14
 Learning Sets of First-Order
Rules: FOIL (Quinlan, 1990)
FOIL is similar to the Propositional Rule learning approach except for the
following:

FOIL accommodates first-order rules and thus needs to accommodate
variables in the rule pre-conditions.

FOIL uses a special performance measure (FOIL-GAIN) which
takes into account the different variable bindings.

FOILS seeks only rules that predict when the target literal is True
(instead of predicting when it is True or when it is False).

FOIL performs a simple hillclimbing search rather than a beam
search.

15
Association Rule Mining (borrowed
from Stan Matwin’s slides)
 Given I={i1,.., im} set of items
 D a set of transaction (a database), each transaction is a set
of items T in 2I .
 Association rule: X => Y, X in I, Y in I, X inter Y = 0
 The support of an itemset is defined as the proportion of
transactions in the data set which contain the itemset.
 The confidence of a rule is defined
conf(X => Y)= supp(X U Y)/supp(X)
 Itemset is frequent if its support > θ

16
 Itemsets and Association Rules
 Itemset = set of items
 k-itemset = set of k-items
 Finding association rules in databases:

Find all frequent (or large) itemsets (those with
support > minS)

Generate rules that satisfy minimum confidence
 Example of an association rule: People who
buy a computer also buy financial software
(support of 2%; confidence of 60%)
17
Example
transacti
milk bread butter beer
on ID
1 1 1 0 0
2 0 0 1 0
3 0 0 0 1
4 1 1 1 0
5 0 1 0 0

 Itemset{milk, bread, butter}


Support 1/5 = .2
 Rule {Bread, Butter} => {Milk}

Confidence = 0.2/0.2 = 1
18
 Apriori Algorithm
 Start with individual items with large support
 In each next step, k,
 Use itemsets from step k-1, generate new itemset Ck
 Compute Ck’s support

Prune the ones that are below the threshold θ

Apriori property: All [non-empty] subsets of a

frequent itemset must be frequent
19
Apriori Algorithm: Example from
Han Kamber, Data Mining, p.232
TID List of Items ID
T100 I1, I2, I5
T200 I2, I4
T300 I2, I3
T400 I1, I2, I4
T500 I1, I3
T600 I2, I3
T700 I1, I3
T800 I1, I2, I3, I5
T900 I1, I2, I3 20
Apriori Algorithm: Example from Han
Kamber, Data Mining, p.232 (Cont’d)

21
From itemsets to association rules

 For each Frequent itemset I


generate all the partitions of I into s, I-s
 Attempt a rule s => I-s iff
support_count(I)/support_count(s) > min c

22