Artificial Intelligence XXXVII: 40th SGAI International Conference on Artificial Intelligence, AI 2020, Cambridge, UK, December 15–17, 2020, Proceedings

Preface
Organization
Contents
Technical Papers
Exposing Students to New Terminologies While Collecting Browsing Search Data (Best Technical Paper)
	1 Introduction
	2 Reviewing Existing Approaches
	3 Proposed Approach
		3.1 Google Chrome Extension
		3.2 Background Server
		3.3 Monitoring Mechanism
	4 Evaluation
		4.1 Computational Experiment Results
		4.2 Student Feedback Evaluation
	5 Conclusion
	References
Neural Nets and Knowledge Management
Symbolic Explanation Module for Fuzzy Cognitive Map-Based Reasoning Models
	1 Introduction
	2 Recurrent Reasoning Module
		2.1 Fuzzy Cognitive Maps
		2.2 Construction and Learning
	3 Symbolic Explanation Module
		3.1 Symbolic Knowledge Representation
		3.2 Generating Explanations
	4 Experiments
	5 Conclusions
	References
Overlap Training to Mitigate Inconsistencies Caused by Image Tiling in CNNs
	1 Introduction
	2 Related Work
	3 Proposed Framework
		3.1 Datasets
		3.2 Applied Convolutional Neural Networks
		3.3 Overlap Loss
	4 Experiments
		4.1 Training Setup
		4.2 Evaluation Metrics
		4.3 Exp1: Can Overlap Training Improve Accuracy, Stability?
		4.4 Exp2: How Overlap Training Affect Boundary Inconsistency?
		4.5 Exp3: Is Overlap Training Better Than Stitching?
	5 Conclusion
	References
The Use of Max-Sat for Optimal Choice of Automated Theory Repairs
	1 Introduction
	2 Background
		2.1 Faults as Reasoning Failures
		2.2 Datalog Theories
		2.3 Deduction by SL Resolution
		2.4  Repair Operations
		2.5 Overproduction of Repair Suggestions
	3 Pruning out Sub-optimal Repairs
		3.1 Turning First-Order Theories into Propositional Logic
		3.2 Partial Max-Sat
		3.3 Evaluating Fitness of Repairs
		3.4 Pareto Optimality
	4 Pruning mechanism
		4.1 The use of Partial MAX-SAT for Determining Optimal Repairs
	5 Evaluation
	6 Conclusion
	References
Machine Learning
Mining Interpretable Rules for Sentiment and Semantic Relation Analysis Using Tsetlin Machines
	1 Introduction
	2 Related Work
	3 Tsetlin Machine for NLP Tasks
		3.1 General Classification and Learning Using Tsetlin Machines
		3.2 Tsetlin Machines in NLP
	4 Experimental Setup
	5 Analysis of Interpretability Provided by Tsetlin Machines
	6 Conclusion
	References
Personalised Meta-Learning for Human Activity Recognition with Few-Data
	1 Introduction
	2 Related Work
	3 Methods
		3.1 Personalised Meta-Learning for HAR
		3.2 Personalised MAML
		3.3 Personalised RN
	4 Evaluation
		4.1 Datasets and Pre-processing
		4.2 Experiment Design
		4.3 Results
	5 Conventional vs. Personalised Meta-Learners
		5.1 MAML vs. MAMLp
		5.2 RN vs. RNp
	6 Discussion
	7 Conclusion
	References
CostNet: An End-to-End Framework for Goal-Directed Reinforcement Learning
	1 Introduction
	2 Background
	3 Related Work
		3.1 Goal-Directed Reinforcement Learning
		3.2 Model-Based Reinforcement Learning
	4 CostNet for Goal-Directed RL
	5 Results and Discussion
		5.1 Results
		5.2 Discussion
	6 Future Work and Conclusion
	References
A Novel Multi-step Finite-State Automaton for Arbitrarily Deterministic Tsetlin Machine Learning
	1 Introduction
	2 A Multi-step Finite-State Learning Automaton
	3 The Arbitrarily Deterministic TM (ADTM)
		3.1 ADTM Inference
		3.2 The MVF-LA Game and Orchestration Scheme
	4 Empirical Evaluation
		4.1 Bankruptcy
		4.2 Balance Scale
		4.3 Breast Cancer
		4.4 Liver Disorders
		4.5 Heart Disease
	5 Effects of Determinism on Energy Consumption
	6 Conclusion
	References
Accelerating the Training of an LP-SVR Over Large Datasets
	1 Introduction
	2 Within-Class Distances for Learning Speed up
		2.1 LP-SVR
		2.2 Definitions
		2.3 Background
		2.4 Within-Class Mahalanobis Distance and Class-Convex Hull
	3 Experimental Results
		3.1 Learning Speedup
	4 Computational Concerns
	5 Conclusion
	References
Short Technical Stream Papers
Learning Categories with Spiking Nets and Spike Timing Dependent Plasticity
	1 Introduction
	2 Literature Review
	3 Methods
	4 Results
	5 Discussion
	References
Developing Ensemble Methods for Detecting Anomalies in Water Level Data
	1 Introduction
	2 Related Works
	3 Water Level Datasets
	4 Anomaly Detection
		4.1 Basic Anomaly Detection Model
		4.2 A Modified Sliding Window Algorithm
		4.3 Simple Ensemble
		4.4 Complex Ensemble
	5 Experiments and Results
	6 Conclusions
	References
Detecting Node Behaviour Changes in Subgraphs
	1 Introduction
	2 Background
	3 Methodology
	4 Experiments
	5 Conclusion
	References
ReLEx: Regularisation for Linear Extrapolation in Neural Networks with Rectified Linear Units
	1 Introduction
	2 Related Work
	3 Model
	4 ReLEx Loss Definitions
	5 Experiments
		5.1 Settings and Metrics
		5.2 Results
	6 Conclusions
	References
Application Papers
Partial-ACO Mutation Strategies to Scale-Up Fleet Optimisation and Improve Air Qualitypg (Best Application Paper)
	1 Introduction
	2 Background and Related Work
	3 Partial-ACO
		3.1 Partial-ACO Applied to Fleet Optimisation
		3.2 Partial-ACO with Mutation for Fleet Optimisation
	4 Results
		4.1 Initial Results
		4.2 Simulating an Ant or Performing a Random Mutation
		4.3 Decoupling Mutation and Ant Simulation
	5 Conclusions
	References
Industrial Applications
A Metaheuristic Search Technique for Solving the Warehouse Stock Management Problem and the Routing Problem in a Real Company
	1 Introduction
	2 Problem Specification
	3 Solving Techniques
		3.1 The Greedy Algorithm
		3.2 GRASP
		3.3 Local Search
		3.4 Improvements
	4 Evaluation
	5 Conclusions and Future Work
	References
Investigating the Use of Machine Learning for South African Edible Garnish Yield Prediction
	1 Introduction
	2 Machine Learning in Agriculture
	3 Details of the Case Study
	4 Exploratory Data Analysis
	5 Empirical Evaluation of the Classifiers
	6 Validation of Variable Importance
	7 Conclusion
	References
Semantic Technologies Towards Accountable Artificial Intelligence: A Poultry Chain Management Use Case
	1 Introduction
	2 Related Work
	3 Accountability of a Predictive Model for a Real-World Poultry Chain
		3.1 Semantic Technologies for Accountability
		3.2 An Implementation in a Real-World Use Case
	4 Discussion and Future Work
	References
Short-Term Forecasting Methodology for Energy Demand in Residential Buildings and the Impact of the COVID-19 Pandemic on Forecasts
	1 Introduction
	2 Short-Term Forecasting Methodology for Time Series
		2.1 Strategies for Multi-Step Ahead Time Series Forecasting
		2.2 KNN Algorithm
		2.3 Evaluation
	3 The RESPOND Energy Demand Forecasting Service
		3.1 The Methodology Implementation
		3.2 The Service Deployment
	4 Evaluation and Results Discussion
		4.1 COVID-19 Impact on Electric Demand Forecasting
	5 Conclusions
		5.1 Future Work
	References
Weather Downtime Prediction in a South African Port Environment
	1 Introduction
	2 Related Literature
	3 Data Description and Data Preparation
		3.1 Data Description
		3.2 Exploratory Data Analysis
		3.3 SOM Cluster Analysis
	4 Regression Algorithm Comparison
	5 Classification Algorithm Comparison
	6 Conclusion
	References
Advances in Applied AI
Software Fault Localisation via Probabilistic Modelling
	1 Introduction
	2 Background
	3 Modelling Program Execution with Naïve Bayes Models
		3.1 Progress Modelling
		3.2 Building the Rich Spectrum
	4 Applying This Model to Fault Localisation
	5 Experiments
		5.1 Experiment 1: Does the Program Progress Measure Actually Measure Progress?
	6 Experiment 2: Can the Program Progress Measure Detect Faults?
	7 Limitations and Threats to Validity
	8 Conclusions and Future Work
	References
Candidates Reduction and Enhanced Sub-Sequence-Based Dynamic Time Warping: A Hybrid Approach
	1 Introduction
	2 Background and Previous Work
	3 Enhanced Sub-Sequence-Based DTW
	4 Time Complexity
	5 Evaluation
		5.1 Parameter Settings
		5.2 Run Time Performance
		5.3 Accuracy of Performance
	6 Conclusion
	References
Ensemble-Based Relationship Discovery in Relational Databases
	1 Introduction
	2 Related Work
	3 Ensemble-Based Discovery
		3.1 Problem Definition
		3.2 Relationship Discovery Algorithms
		3.3 Ensemble Strategies
	4 Experimental Evaluation
		4.1 Dataset Description
		4.2 Experimental Set-Up
		4.3 Evaluation Metrics
		4.4 Comparative Analysis
	5 Conclusion
	References
Intention-Aware Model to Support Agent Deliberation in a Large-Scale Dynamic Multi-Agent Application
	1 Introduction
	2 Related Models
		2.1 Route Selection Models
		2.2 Congestion Sensitivity Models
		2.3 Prediction Models
	3 Intention-Aware Model for Real-World Traffic Routing
	4 Implementation and Test Installation of the Intention-Aware Model
	5 Real-World Test Scenario
	6 Performance in the Test Scenario
	7 Conclusion
	References
Medical and Legal Applications
Combining Bandits and Lexical Analysis for Document Retrieval in a Juridical Corpora
	1 Introduction
	2 Corpus and Search Interface
	3 Background
		3.1 Definitions
		3.2 Lexical Pertinence
		3.3 Similarity Metrics
		3.4 User Feedback
	4 Learning-to-Rank
	5 Proposed Model
	6 Conclusions
	References
In-Bed Human Pose Classification Using Sparse Inertial Signals
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Virtual Human Motion Capture
		3.2 Articulated Body Representation
		3.3 Classification of Sleeping Poses
	4 Experiments and Discussion of Results
	5 Conclusions
	References
Maintaining Curated Document Databases Using a Learning to Rank Model: The ORRCA Experience
	1 Introduction
	2 Literature Review
	3 The Curated Document Database Update Approach
	4 Pre-processing and Feature Extraction
	5 Evaluation
	6 Conclusion
	References
What Are We Depressed About When We Talk About COVID-19: Mental Health Analysis on Tweets Using Natural Language Processing
	1 Introduction
	2 Dataset
	3 Classification
	4 Correlation
		4.1 Attention Weight
		4.2 POS Tagging
	5 Emotion Trend Analysis
		5.1 Emotion Trend with Keywords
		5.2 Emotion Trend with Topics
	6 Conclusion and Future Work
	References
Short Application Stream Papers
Using Sentence Embedding for Cross-Language Plagiarism Detection
	1 Introduction
	2 Proposed Method
		2.1 Sentence Embedding Based on CL-WE-Tw Model
		2.2 Sentence Embedding Based on MUSE
		2.3 Overall Sentence Similarity
	3 Experiment and Results
	4 Conclusion
	References
Leveraging Anomaly Detection for Proactive Application Monitoring
	1 Introduction
	2 Related Works
	3 Proposed Approach
		3.1 Outlier Detection
		3.2 Novelty Detection
		3.3 Implementation and Model Training
	4 Experiment Results and Application
	5 Conclusion
	References
Using Active Learning to Understand the Videoconference Experience: A Case Study
	1 Introduction
	2 Challenges Arising
	3 Proposed Approach
		3.1 Active Learning
		3.2 Data Features
	4 Outcomes
	5 Conclusion
	References
An Application of EDA and GA for Permutation Based Spare Part Allocation Problem
	1 Introduction
	2 Background and Problem Formulation
	3 EA Approach to Solve the Problem
	4 Experimental Results
	5 Conclusion and Future Work
	References
Do You Remember Me? Betty the Conversational Agent
	1 Introduction
		1.1 Ageing Population Growth
		1.2 Conversational Agent Framework
	2 Related Work
	3 Developing Betty
		3.1 Components of the System
		3.2 The System in Operation
	4 Conclusions
	References
Author Index