Inductive Logic Programming: 30th International Conference, ILP 2021, Virtual Event, October 25–27, 2021, Proceedings

✍ Scribed by Nikos Katzouris, Alexander Artikis

• Publisher: Springer
• Year: 2022
• Tongue: English
• Leaves: 293
• Series: Lecture Notes in Computer Science, 13191
• Category: Library

✦ Synopsis

This book constitutes the refereed conference proceedings of the 30th International Conference on Inductive Logic Programming, ILP 2021, held in October 2021. Due to COVID-19 pandemic the conference was held virtually.

The 16 papers and 3 short papers presented were carefully reviewed and selected from 19 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
Embedding Models for Knowledge Graphs Induced by Clusters of Relations and Background Knowledge*-6pt
1 Introduction
2 Basics on KGs and Embedding Models
3 TransROWL-HRS
3.1 TransROWL and Relation Hierarchies
3.2 Model Formalization
3.3 Training the Model
4 Empirical Evaluation
4.1 Experiment Setup
4.2 Link Prediction
4.3 Triple Classification
5 Related Work
6 Conclusions and Future Work
References
Automatic Conjecturing of P-Recursions Using Lifted Inference
1 Introduction
2 Background
2.1 First-Order Logic and Model Counting
2.2 P-Recursive Sequences
3 Approach
3.1 Conjecturing Recurrence Relations
4 Experiments
5 Conclusions
References
Machine Learning of Microbial Interactions Using Abductive ILP and Hypothesis Frequency/Compression Estimation
1 Introduction
2 Background and Related Work
3 Methods
3.1 Logical Description of Microbial Interactions
3.2 Bootstrapping
3.3 Simulated Data-Sets
3.4 Compositionality and Bias
4 Experimental Evaluation
4.1 Experiment 1
4.2 Experiment 2
5 Discussion and Conclusion
References
Answer-Set Programs for Reasoning About Counterfactual Interventions and Responsibility Scores for Classification
1 Introduction
2 Counterfactual Interventions and Explanation Scores
3 A Naive-Bayes Classifier
4 The x-Resp Score
5 Counterfactual-Intervention Programs
6 Exploiting Domain Knowledge and Query Answering
7 Final Remarks
References
Synthetic Datasets and Evaluation Tools for Inductive Neural Reasoning
1 Introduction
2 Rule Learning Preliminaries
3 RuDaS Datasets
4 Evaluation Tools
5 Experiments
5.1 Overall Results: Comparing Different Metrics
5.2 Impact of Missing Consequences and Noise
5.3 Impact of Dependencies Between Rules
5.4 Scalability: Impact of Dataset Size
6 Conclusions and Future Work
A Additional Preliminaries
B Dataset Generation
C Statistics of RuDaS.v0
D Approaches to Rule Learning
E System Configurations
References
Using Domain-Knowledge to Assist Lead Discovery in Early-Stage Drug Design
1 Introduction
2 System Design and Implementation
3 System Testing
3.1 Materials
3.2 Methods
3.3 Results
4 Related Work
5 Concluding Remarks
A Domain-Knowledge Used in Experiments
B Proxy Model for Predicting Hit Confirmation
References
Non-parametric Learning of Embeddings for Relational Data Using Gaifman Locality Theorem
1 Introduction
2 Background and Related Work
3 Gaifman-Guided Learning of Relational Embeddings
4 Experiments
5 Conclusion and Future Work
References
Ontology Graph Embeddings and ILP for Financial Forecasting
1 Introduction
2 Background
2.1 Financial Forecasting
2.2 Pair Trading
2.3 Securities and Exchange Commission (SEC) Reports
2.4 Ontologies
2.5 Node2vec Embeddings
3 Methodology
3.1 Data Collection
3.2 Data Analysis
3.3 Learning and Experiment Preparation
4 Experiment Design and Results
5 Conclusion
References
Transfer Learning for Boosted Relational Dependency Networks Through Genetic Algorithm
1 Introduction
2 Background Knowledge
2.1 RDN-Boost
2.2 Transfer Learning
2.3 Genetic Algorithm
3 Related Work
4 GROOT: Genetic Algorithms to Aid Transfer Learning with bOOsTsrl
4.1 Population
4.2 Genetic Operators
4.3 Selection and Evaluation
5 Experimental Results
5.1 Datasets
6 Conclusion
References
Online Learning of Logic Based Neural Network Structures
1 Introduction
2 Background Knowledge
3 Online Theory Revision with NeuralLog
3.1 Online Structure Learner by Revision
3.2 Online Meta-Interpretive Learning
4 Experiments
4.1 Datasets
4.2 Simulating the Online Environment
4.3 Results
5 Related Work
6 Conclusion
References
Programmatic Policy Extraction by Iterative Local Search
1 Introduction
2 Methods
2.1 Program Synthesis by Type-Directed Search
2.2 Typed Neighborhood
2.3 Policy Extraction by Local Synthesis
3 Experiments
3.1 Programming by Example with Local Program Search
3.2 Imitation of a Programmatic Pendulum Swing-Up Policy
3.3 Imitation of a Neural Pendulum Swing-Up Policy
4 Discussion
4.1 Related Work
References
Mapping Across Relational Domains for Transfer Learning with Word Embeddings-Based Similarity
1 Introduction
2 Background Knowledge
2.1 Functional Gradient Boosting of Relational Dependency Networks
2.2 Transfer Learning
2.3 Word Embeddings
3 Related Work
4 TransBoostler
4.1 Transferring and Revising the Structure
4.2 Mapping Component
5 Experimental Results
6 Conclusions
References
A First Step Towards Even More Sparse Encodings of Probability Distributions
1 Introduction
2 Notations and Problem Statement
3 CoFE: Compact Formula Extraction
4 Empirical Evaluation
4.1 Test Setting
4.2 Distance and Reconstruction
4.3 Sparsity and Error
5 Conclusion
References
Feature Learning by Least Generalization
1 Introduction
2 Feature Learning by Least Generalization
2.1 Least Generalization
2.2 Encoding Image Data into First-Order Expressions
2.3 Extracting Features
2.4 Classification of Images
3 Experimental Results
3.1 Testing on MNIST Dataset
3.2 Testing on Fashion-MNIST Dataset
4 Discussion
5 Concluding Remarks
References
Learning Logic Programs Using Neural Networks by Exploiting Symbolic Invariance
1 Introduction
2 Background
2.1 LFIT
2.2 Rule Classification
2.3 LFIT
3 LFIT+
3.1 Input Sequence Invariance
3.2 Rule Length Sharing
3.3 Subsumed Label Smoothing
3.4 Label Imbalance
3.5 Network Architecture
3.6 Training Methods
3.7 LFIT+ Algorithm
4 Experiments
4.1 Experimental Methods
4.2 Baseline
4.3 Regular Transformer
4.4 Rule Length Sharing
4.5 Without Label Smoothing
5 Discussion
6 Related Work
7 Conclusion
References
Learning and Revising Dynamic Temporal Theories in the Full Discrete Event Calculus
1 Introduction
2 Background and Related Work
3 The eXploratory Event Calculus (XEC)
4 XEC Theory Learning and Revision with XHAIL
5 Conclusion
References
Human-Like Rule Learning from Images Using One-Shot Hypothesis Derivation
1 Introduction
2 One-Shot Hypothesis Derivation (OSHD)
3 Siamese Neural Networks
4 One-Shot Learning for Malayalam Character Recognition
4.1 Character Recognition and Human-Like Background Knowledge
4.2 Mode Declarations
5 One-Shot Learning for Neurological Diagnosis Using Retinal Images
5.1 Retinal Vasculature Features
5.2 Retinal Fundus Image Dataset
5.3 Extraction of Subject Data and Background Knowledge (BK) Preparation
6 Experiments
6.1 Materials and Methods
6.2 Results and Discussions
7 Conclusion
References
Generative Clausal Networks: Relational Decision Trees as Probabilistic Circuits
1 Introduction
2 Background and Related Work
3 Probabilistic Circuits over Logical Clauses
4 Experiments
5 Conclusion
References
A Simulated Annealing Meta-heuristic for Concept Learning in Description Logics
1 Introduction
2 Related Work
3 Approach
3.1 Notation and Preliminaries
3.2 Concept Learning in Description Logics
3.3 Simulated Annealing
3.4 Simulated Annealing in Concept Learning
4 Empirical Evaluation
4.1 Evaluation Setup
4.2 Results
5 Conclusions
References
Author Index