Folie 1

1
Application of Probabilistic ILP II, FP6-508861
www.aprill.org
Probabilistic Logic Learning
al and Relational
Probability
Logic
Learning
James Cussens University of York UK
Kristian Kersting University of Freiburg Germany
2
Special thanks to the APrIL II consortium

• Application of Probabilistic ILP
• 3 years EU project
• 5 institutes
• www.aprill.org

Heikki Mannila
Stephen Muggleton, Mike Sternberg Subcontractor
James Cussens
Luc De Raedt Subcontractor Manfred Jaeger
François Fages
Paolo Frasconi
3
... special thanks ...

• ... for discussions, materials, and
  collaborations to
• Alexandru Cocura, Uwe Dick, Pedro Domingos,
  Peter Flach, Thomas Gaertner, Lise Getoor, Martin
  Guetlein,
• Bernd Gutmann, Tapani Raiko, Reimund Renner,
  Richard Schmidt, Ingo Thon, ...

4
Tutorials Aims

• Introductory survey
• Identification of important probabilistic,
  relational/logical and learning concepts

5
Objectives
One of the key open questions of AI concerns
Probabilistic Logic Learning
The integration of probabilistic reasoning with
Probabilitiy
first order / relational logic representations and
Logic
Learning
machine learning.
6
Why do we need PLL?
Robotics
Medicine
Diagnosis Prediction Classification Decision-makin
g Description
Web Mining
Computational Biology
Lets look at an example
PLMs
7
Web Mining / Linked Bibliographic Data /
Recommendation Systems /
illustration inspired by Lise Getoor
book
book
author
book
author
publisher
book
publisher
Real World
8
Web Mining / Linked Bibliographic Data /
Recommendation Systems /
books
B2
authors
publishers
B1
B3
series
A2
author-of
publisher-of
A1
P2
B4
P1
Fantasy
Science Fiction
Real World
9
Why do we need PLL?
Real World Applications
Lets look at some more examples
10
Blood Type / Genetics/ Breeding

• 2 Alleles A and a
• Probability of Genotypes AA, Aa, aa ?

Father
Mother
Offspring
Prior for founders
CEPH Genotype DB,http//www.cephb.fr/
11
Bongards Problems
Noise?
Some objects are opaque? (e.g. in relation is not
always observed)
12
Bongards Problems
Noise?
Some objects are opaque? (e.g. in relation is not
always observed)
Clustering?
13
Others
Social Networks
Protein Secondary Structure
Data Cleaning
Scene interpretation
?
Metabolic Pathways
Phylogenetic Trees
14
Why do we need PLL ?
Statistical Learning (SL)
Probabilistic Logics
Uncertainty

• attribute-value representations some learning
  problems cannot (elegantly) be described using
  attribute value representations

• no learning to expensive to handcraft models

soft reasoning, expressivity
soft reasoning, learning
PLL
Real World Applications
Structured Domains
Machine Learning
Inductive Logic Programming (ILP) Multi-Relational
Data Mining (MRDM)
- crisp reasoning some learning problems
cannot (elegantly) be described without explicit
handling of uncertainty
expressivity, learning
15
Why do we need PLL?

• Rich Probabilistic Models
• Comprehensibility
• Generalization (similar situations/individuals)
• Knowledge sharing
• Parameter Reduction / Compression
• Learning
• Reuse of experience (training one RV might
  improve prediction at other RV)
• More robust
• Speed-up

16
When to apply PLL ?

• When it is impossible to elegantly represent your
  problem in attribute value form
• variable number of objects in examples
• relations among objects are important
• Background knowledge can be defined intensionally
  
• define benzene rings as view predicates

17
Why stressing Learning?
Learning Algorithm
Database
Model

• Knowledge acquisition bottleneck / data is cheap
• General purpose systems
• Combining domain expert knowledge with data
• Logical structure provides insight into domain
• Handling missing data bt(luc)?

18
(Incomplete) Historical Sketch
names in alphabetical order
99
03
Present
Future
2003
many more ...
19
Overview

• Introduction to PLL
• Foundations of PLL
• Logic Programming, Bayesian Networks, Hidden
  Markov Models, Stochastic Grammars
• Frameworks of PLL
• Independent Choice Logic,Stochastic Logic
  Programs, PRISM,
• Bayesian Logic Programs, Probabilistic Logic
  Programs,Probabilistic Relational Models
• Logical Hidden Markov Models
• Applications