Inductive Logic Programming: 27th International Conference, ILP 2017, Orléans, France, September 4-6, 2017, Revised Selected Papers (Lecture Notes in Computer Science, 10759)

✍ Scribed by Nicolas Lachiche (editor), Christel Vrain (editor)

• Publisher: Springer
• Year: 2018
• Tongue: English
• Leaves: 195
• Category: Library

✦ Synopsis

This book constitutes the thoroughly refereed post-conference proceedings of the 27th International Conference on Inductive Logic Programming, ILP 2017, held in Orléans, France, in September 2017.
The 12 full papers presented were carefully reviewed and selected from numerous submissions.
Inductive Logic Programming (ILP) is a subfield of machine learning, which originally relied on logic programming as a uniform representation language for expressing examples, background knowledge and hypotheses. Due to its strong representation formalism, based on first-order logic, ILP provides an excellent means for multi-relational learning and data mining, and more generally for learning from structured data.

✦ Table of Contents

Preface
Organization
Contents
Relational Affordance Learning for Task-Dependent Robot Grasping
1 Introduction
2 Related Work
3 Problem Description and Representation
3.1 Object Category-Task (CT) Affordances and Constraints
3.2 Object Part-Category-Task (PCT) Affordances
3.3 The Affordance Learning Problem
4 Approach: Probabilistic Rule Learning
5 Experiments
5.1 Results for Object Category-Task (CT) Affordances
5.2 Results for Object Category-Task (CT) Constraints
5.3 Results for Object Part-Category-Task (PCT) Affordances
5.4 Discussion
6 Conclusions
References
Positive and Unlabeled Relational Classification Through Label Frequency Estimation
1 Introduction
2 Related Work
3 PU Learning and the Label Frequency
3.1 Positive Unlabeled (PU) Learning
3.2 Using the Label Frequency to Simplify PU Learning
4 Label Frequency Estimation
4.1 Label Frequency Estimation in Propositional PU Data
4.2 Label Frequency Estimation in Relational PU Data
5 Experiments
5.1 Datasets
5.2 Methods
5.3 c-adjusted TILDE: Performance and Sensitivity to
5.4 c-adjusted Aleph: Performance and Sensitivity to
5.5 TIcER Evaluation
5.6 Method Comparison
6 Conclusions
References
On Applying Probabilistic Logic Programming to Breast Cancer Data
1 Introduction
2 Probabilistic Inductive Logic Programming
3 Methodology
4 Experiments
5 Related Work
6 Conclusion
References
Logical Vision: One-Shot Meta-Interpretive Learning from Real Images
1 Introduction
2 Related Work
3 Framework
3.1 Meta-Interpretive Learning
3.2 Logical Vision
4 Implementation
4.1 Meta-Interpretation in Real Images
5 Experiments
5.1 Materials
5.2 Methods
5.3 Results and Discussion
6 Using LV for Secondary Reasoning: Interpreting Ambiguity
7 Conclusions and Further Work
References
Demystifying Relational Latent Representations
1 Introduction
2 Background
2.1 Neighbourhood Trees
2.2 CUR2LED
3 Opening the Black Box of Latent Features
3.1 Extracting the Meaning of Latent Features
3.2 Properties of Latent Spaces
4 Experiments and Results
4.1 Datasets and Setup
4.2 Interpretability
4.3 Properties of Latent Spaces
4.4 Looking Forward
5 Conclusion
References
Parallel Online Learning of Event Definitions
1 Introduction
2 Background
3 The OLED System
4 A Parallel Version of OLED
4.1 Main Operations of the Parallel OLED Strategy
4.2 Decentralized Coordination
5 Empirical Evaluation
6 Related Work
7 Conclusions and Future Work
References
Relational Restricted Boltzmann Machines: A Probabilistic Logic Learning Approach
1 Introduction
2 Prior Work and Background
2.1 Random Walks
2.2 Relational Probabilistic Models
2.3 Structure Learning Approaches
2.4 Propositionalization Approaches
2.5 Restricted Boltzmann Machines
3 Relational Restricted Boltzmann Machines
3.1 Step 1: Relational Data Representation
3.2 Step 2: Relational Transformation Layer
3.3 Step 3: Learning Relational RBMs
3.4 Relation to Probabilistic Relational Models
4 Experiments
4.1 Data Sets
4.2 Results
5 Conclusion
References
Parallel Inductive Logic Programming System for Superlinear Speedup
1 Introduction
2 Improved ILP System
3 Improvement of Parallel ILP System
3.1 Previous Parallel ILP System
3.2 Improvement of the Parallel ILP System Communication Protocol
4 Experiment and Results
4.1 Experiment Problems
4.2 Experiment Results
5 Discussion
5.1 Discussion of Speedup
5.2 Discussion of Speedup Stability
6 Conclusions
References
Inductive Learning from State Transitions over Continuous Domains
1 Introduction
2 Continuum Logic and Program Learning
2.1 Continuum Logic Programs
2.2 Learning Operations
3 ACEDIA
3.1 Algorithm
3.2 Evaluation
4 Conclusions
References
Stacked Structure Learning for Lifted Relational Neural Networks
1 Introduction
2 Preliminaries
3 Structure Learning
3.1 Structure of the Learned LRNNs
3.2 Structure Learning Algorithm
4 Experiments
4.1 Molecular Datasets
4.2 A Hard Artificial Problem
5 Related Work
6 Conclusions and Future Work
References
Pruning Hypothesis Spaces Using Learned Domain Theories
1 Introduction
2 Preliminaries
2.1 First-Order Logic
2.2 Learning Setting
2.3 Theorem Proving Using SAT Solvers
3 Pruning Hypothesis Spaces Using Domain Theories
3.1 Saturations
3.2 Searching the Space of Saturations
3.3 Pruning Isomorphic Saturations
3.4 Learning Domain Theories for Pruning
3.5 Why Relative Subsumption is Not Sufficient
4 Experiments
4.1 Methodology and Implementation
4.2 Results
5 Related Work
6 Conclusions
References
An Investigation into the Role of Domain-Knowledge on the Use of Embeddings
1 Introduction
2 Baseline and Vector-Space Representations
2.1 First-Order Feature-Based Representation
2.2 Vector-Space Representation
3 Empirical Evaluation
3.1 Aims
3.2 Materials
3.3 Method
3.4 Results
4 Concluding Remarks
References
Author Index