Inductive Logic Programming: 17th International Conference, ILP 2007, Corvallis, OR, USA, June 19-21, 2007, Revised Selected Papers (Lecture Notes in Computer Science, 4894)

✍ Scribed by Hendrik Blockeel (editor), Jan Ramon (editor), Jude Shavlik (editor), Prasad Tadepalli (editor)

• Publisher: Springer
• Year: 2008
• Tongue: English
• Leaves: 318
• Category: Library

✦ Synopsis

This book constitutes the thoroughly refereed post-conference proceedings of the 17th International Conference on Inductive Logic Programming, ILP 2007, held in Corvallis, OR, USA, in June 2007 in conjunction with ICML 2007, the International Conference on Machine Learning. The 15 revised full papers and 11 revised short papers presented together with 2 invited lectures were carefully reviewed and selected from 38 initial submissions. The papers present original results on all aspects of learning in logic, as well as multi-relational learning and data mining, statistical relational learning, graph and tree mining, relational reinforcement learning, and learning in other non-propositional knowledge representation frameworks. Thus all current topics in inductive logic programming, ranging from theoretical and methodological issues to advanced applications in various areas are covered.

✦ Table of Contents

Title Page
Preface
Organization
Table of Contents
Learning with Kernels and Logical Representations
Motivations
Overview of Methods
Beyond Prediction: Directions for Probabilistic and Relational Learning
Introduction
Why Causal Models Are Useful
Example: Fraud Detection at the NASD
NASD
Available Data
The Potential for Causal Models
Current Practice
Experimental Design
Joint Modeling
Quasi-experimental Design
Limitations of Current Approaches
Research Opportunities
Timeliness
Risks and Benefits
References
Learning Probabilistic Logic Models from Probabilistic Examples (Extended Abstract)
Learning Directed Probabilistic Logical Models Using Ordering-Search
Learning to Assign Degrees of Belief in Relational Domains
Bias/Variance Analysis for Relational Domains
Induction of Optimal Semantic Semi-distances for Clausal Knowledge Bases
Introduction
A Family of Semi-distances for Instances
Basic Measure Definition
Discussion
Extensions
Optimization
Experiments on Similarity Search
Conclusions and Ongoing Work
Clustering Relational Data Based on Randomized Propositionalization
Introduction
Method
Experiments
Summary and Future Work
Structural Statistical Software Testing with Active Learning in a Graph
Introduction
Position of the Problem
Statistical Structural Software Testing
SSST and Supervised Learning
Extended Parikh Representation
Overview
Principle
Init Module
Constrained Exploration Module
Generalization Module
Experimental Validation
Experimental Setting
Experimental Results
Discussion
Conclusion and Perspectives
Learning Declarative Bias
Introduction
Inductive Process Modeling
Learning Bias by Inductive Logic Programming
Empirical Evaluation
Method
Selecting a Performance Threshold
Evaluating the Generalization Performance
Semantic Analysis of the Induced Constraints
General Discussion
Conclusion
ILP :- Just Trie It
Introduction
The Trie Data Structure
Trieing MDIE
Trie the Real World
Experiments and Results
Conclusions
Learning Relational Options for Inductive Transfer in Relational Reinforcement Learning
Introduction
Background
Relational Reinforcement Learning and the Blocks World
The Options Framework
Transfer Learning
Relational Options
Relational Skill Learning
Experimental Evaluation
Conclusions and Further Work
Empirical Comparison of “Hard” and “Soft” Label Propagation for Relational Classification
Introduction
Problem Settings
Algorithms
Score Propagation
Label Propagation
Related Work
Experiments with Synthetic Data
Class Overlap
ROC Analysis
Effect of Noise
Experiments with the CoRA Data
Discussion and Future Work
A Phase Transition-Based Perspective on Multiple Instance Kernels
Introduction
State of the Art
Overview
When MI Learning Meets Linear Programming
Order Parameters and Experimental Setting
Goal of the Experiments
Experiments
Summary of the Results
LPP Satisfiability Landscape
Generalization Error Landscape
Conclusion and Perspectives
Combining Clauses with Various Precisions and Recalls to Produce Accurate Probabilistic Estimates
Introduction
Learning Probabilities with GleanerSRL
Gleaning Clauses
Creating Features
Learning to Predict Scores
Calibrating Probabilities
Experimental Results
Related and Future Work
Applying Inductive Logic Programming to Process Mining
Introduction
A Representation for Process Traces and Models
Learning ICs Theories
Experiments
Related Works
Conclusions and Future Works
A Refinement Operator Based Learning Algorithm for the ALC Description Logic
Introduction
Preliminaries
Description Logics
Learning in Description Logics Using Refinement Operators
Designing a Refinement Operator
Completeness of the Operator
Achieving Properness
The Learning Algorithm
Redundancy Elimination
Creating a Full Learning Algorithm
Preliminary Evaluation
Related Work
Conclusions and Further Work
Foundations of Refinement Operators for Description Logics
Introduction
Description Logics
Learning in Description Logics Using Refinement Operators
Analysing the Properties of Refinement Operators
Related Work
Conclusions
A Relational Hierarchical Model for Decision-Theoretic Assistance
Introduction
Decision-Theoretic Assistance
A Relational Hierarchical Model of Assistance
Relational Hierarchical Policies
Goal Estimation
Action Selection
Experiments and Results
Doorman Domain
Kitchen Domain
Related Work
Conclusions and Future Work
Using Bayesian Networks to Direct Stochastic Search in Inductive Logic Programming
Introduction
Directed Stochastic Search Algorithm
Modeling ILP's Search Space with Bayesian Networks
Training the Model
Using the Model to Guide Search
Directed-Search Experiments
Related Work
Conclusions and Future Work
Revising First-Order Logic Theories from Examples Through Stochastic Local Search
Introduction
Stochastic Search
First-Order Logic Theory Revision
Stochastic First-Order Logic Theory Revision
Stochastic Local Search for Antecedents
Stochastic Local Search for Revisions
Experimental Results
Conclusions
Using ILP to Construct Features for Information Extraction from Semi-structured Text
Introduction
Feature Definitions Using Inductive Logic Programming
Experimental Evaluation
Aims
Materials
Method
Results
Concluding Remarks
Mode-Directed Inverse Entailment for Full Clausal Theories
Introduction
Background
Notation and Terminology
Mode Declarations
MDIE (Mode Directed Inverse Entailment)
HAIL (Hybrid Abductive Inductive Learning)
SOLAR (SOL Resolution for Advanced Reasoning)
Motivating Example: Fluid Modelling
Full Clausal Hybrid Abductive Inductive Learning
Related Work
Conclusions
Mining of Frequent Block Preserving Outerplanar Graph Structured Patterns
Introduction
Graph Pattern
Block Preserving Outerplanar Graph Patterns and Block Tree Patterns
Matching Algorithm for Block Preserving Outerplanar Graph Patterns
Pattern Enumeration Algorithm for Frequent BPO Graph Pattern Problem
Experimental Result
Conclusion and Future Works
Relational Macros for Transfer in Reinforcement Learning
Introduction
Reinforcement Learning in RoboCup
Related Work in Transfer Learning
Executing a Relational Macro
Learning a Relational Macro
Structure Learning
Ruleset Learning
Transferring a Relational Macro
Experimental Results
Conclusions and Future Work
Seeing the Forest Through the Trees Learning a Comprehensible Model from a First Order Ensemble
Introduction
Proposed Method
Computing Heuristics from the Ensemble
Generation of Candidate Test Queries
Computing the Optimal Split
Stop Criteria
Prediction of a Leaf in the New Tree
Empirical Evaluation
Conclusions and Future Work
Building Relational World Models for Reinforcement Learning
Introduction
Background
Building World Models
Terminology
Algorithm Overview
Preimage Selection
Learning Concepts Via ILP
Building the MDP
The RL Learning Cycle
Empirical Results
Domains
Learning Algorithms
Results
Related Work
Conclusions and Future Work
References
An Inductive Learning System for XML Documents
Introduction
Knowledge Representation for XML Documents
The Structure of an XML Document
Representation of Individuals
Representation of Features
Precision/Recall-Driven Decision-Tree (PRDT) Algorithm
Precision and Recall
Structured Feature Selection
Node Selection
The Precision/Recall-Driven Decision-Tree Algorithm
Experiments
The Dataset
Experimental Results
Conclusion
Author Index