Abbildung von: Artificial Intelligence XXXVI - Springer

Artificial Intelligence XXXVI

39th SGAI International Conference on Artificial Intelligence, AI 2019, Cambridge, UK, December 17-19, 2019, Proceedings

Max Bramer Miltos Petridis(Herausgeber*in)

Springer (Verlag)

Erschienen am 9. Dezember 2019

XV, 502 Seiten

E-Book

PDF mit Wasserzeichen-DRM

Systemvoraussetzungen

978-3-030-34885-4 (ISBN)

53,49 €inkl. 7% MwSt.

Systemvoraussetzungen

für PDF mit Wasserzeichen-DRM

E-Book Einzellizenz

Als Download verfügbar

Beschreibung

Weitere Details

Weitere Ausgaben

Inhalt

Intro
Preface
Organization
Contents
Technical Papers
CascadeML: An Automatic Neural Network Architecture Evolution and Training Algorithm for Multi-label Classification (Best Technical Paper)
1 Introduction
2 Related Work
2.1 Neural Networks for Multi-label Classification and BPMLL
2.2 The Cascade2 Algorithm
3 The CascadeML Algorithm
4 Experiment Design
5 Results
6 Conclusions and Future Work
References
Machine Learning, Knowledge Discovery and Data Mining
Purity Filtering: An Instance Selection Method for Support Vector Machines
1 Introduction
2 Relevant Previous Work
3 Purity Filtering
3.1 Data Reduction
3.2 Hyperplane Estimation
3.3 Margin Estimation
3.4 Data Selection
3.5 Theoretical Analysis
4 Experimentation and Discussion
4.1 Synthetic Data
4.2 Covertype Dataset
5 Conclusions
References
Towards Model-Based Reinforcement Learning for Industry-Near Environments
1 Introduction
2 Literature Review
2.1 Automated Storage and Retrieval Systems (ASRS)
2.2 Model-Based Reinforcement Learning
3 Background
4 Learning Policies Using Predictive Models
4.1 Motivation and Environment Safety
4.2 The Dreaming Variational Autoencoder V2
4.3 Model Selection
4.4 Implementation
5 The Deep Warehouse Environment
5.1 Motivation
5.2 Implementation
6 Experimental Results
6.1 The Importance of Compute
6.2 Results
7 Conclusion and Future Work
References
Stepwise Evolutionary Learning Using Deep Learned Guidance Functions
1 Introduction
2 Deep Learned Guidance Functions
2.1 Formalisation
3 Example: Applying Deep LGFs to the Rubik's Cube
3.1 Constructing the LGF
3.2 Sources of Error
4 Experiments and Results
5 Related Work
6 Summary and Future Work
References
Monotonicity Detection and Enforcement in Longitudinal Classification
1 Introduction
2 Background
2.1 Longitudinal Classification
2.2 Monotonic Classification
2.3 Monotonicity Measures
2.4 The XGBoost Tree Boosting Algorithm
3 Methodology
3.1 Quantitative Measures for Monotonicity Detection
3.2 The Proposed Longitudinal Monotonicity Detection Approaches
3.3 Datasets
3.4 Experimental Setup
4 Results of the Experiments and Comparative Analysis
4.1 Predictive Accuracies of Constructed Models
4.2 Monotonicity Constraints and Their Effects on Model Sizes
4.3 Feature Importance
5 Conclusion
References
Understanding Structure of Concurrent Actions
1 Introduction
2 Related Work
3 Preliminaries
4 Implicit Structure: Action Elimination Using Task Invariance
5 Explicit Structure: Clustering of Action Space
6 Experiments
6.1 Implicit Structure - Action Elimination
6.2 Explicit Structure - Clustering
7 Conclusion and Future Work
References
Agents, Knowledge Acquisition and Ontologies
Demonstrating the Distinctions Between Persuasion and Deliberation Dialogues
1 Introduction
2 Speech Acts
2.1 Refinement of Speech Acts
3 Design
3.1 Agent Class
3.2 Dialogue Class
3.3 PersuasionDialogue Class
3.4 DeliberationDialogue Class
3.5 Simulation Class
4 Realisation
4.1 Project Architecture
4.2 Interface
5 Description of the Tool
6 Testing and Evaluation
6.1 Unit Testing
7 Discussion and Concluding Remarks
References
Ontology-Driven, Adaptive, Medical Questionnaires for Patients with Mild Learning Disabilities
Abstract
1 Introduction
2 Background and Related Work
3 Methodology
4 Ontologies Development
4.1 Medical Questionnaire Ontology
4.2 Accessibility Ontology
5 System Implementation - Java Adaptive Engine
5.1 Java Engine
5.2 Dynamic Changes to Stack
6 Scenario Based Evaluation
7 Conclusion and Future Work
References
Exposing Knowledge: Providing a Real-Time View of the Domain Under Study for Students
1 Introduction
2 Current Solutions
3 Proposed Knowledge Search Framework
3.1 Background Worker
3.2 User Interface
3.3 Similarity Analysis
4 Evaluation
5 Conclusion
References
Short Technical Papers
A General Approach to Exploit Model Predictive Control for Guiding Automated Planning Search in Hybrid Domains
1 Introduction
2 Background
3 Using MPC to Guide Automated Planning Search
4 Experimental Analysis
5 Conclusion
References
A Tsetlin Machine with Multigranular Clauses
1 Introduction
2 A Tsetlin Machine with Multigranular Clauses
3 Experimental Results
4 Conclusion
References
Building Knowledge Intensive Architectures for Heterogeneous NLP Workflows
Abstract
1 Introduction
2 Related Work
3 An Architecture for Knowledge Intensive Workflows
4 DeepKAF Evaluation
5 Conclusions
References
WVD: A New Synthetic Dataset for Video-Based Violence Detection
1 Introduction
2 Hot vs Cold Weapons Violence Dataset
3 Experimental Evaluation
4 Conclusion and Future Work
References
Application Papers
Evolving Prediction Models with Genetic Algorithm to Forecast Vehicle Volume in a Service Station (Best Application Paper)
Abstract
1 Introduction
2 Background
3 Model Formulation
3.1 Neural Network Based Prediction Model
3.2 Genetic Algorithm
4 Experiment Setup and Analysis of Results
4.1 Comparison of Accuracy Using MAPE
4.2 Comparison of Produced Parameter Settings
4.3 Typical GA Evolution of Model
5 Conclusion
References
Medical Applications
Are You in Pain? Predicting Pain and Stiffness from Wearable Sensor Activity Data
1 Introduction
2 Related Work
3 Participants and Data Acquisition
3.1 Participants and Recruitment Procedure
3.2 Data Collection
4 Transformation of Actigraph Data
4.1 Generating 60s Epoch Measures
5 Generating Actigraphy Variables from 60s Epoch Measures
5.1 Intra-DAD Measures
5.2 Inter-DAD Measures
6 Predicting Morning Pain and Stiffness
6.1 Feature Selection and Modelling Algorithms
6.2 Evaluation Method
6.3 Results
7 Conclusions and Future Work
References
Motif Discovery in Long Time Series: Classifying Phonocardiograms
1 Introduction
2 Previous Work
3 PCG Frequent Motif Selection and Extraction
3.1 Cycle Segmentation
3.2 Candidate Motif Selection
3.3 Frequent Motif Extraction
4 Evaluation
4.1 Evaluation Data
4.2 Experimental Set-Up
4.3 Cycle Segmentation Subprocess Evaluation
4.4 Candidate Motif Selection Subprocess Evaluation
4.5 Frequent Motif Extraction Subprocess Evaluation
4.6 Runtime Evaluation
4.7 Classification Accuracy
5 Conclusions
References
Exploring the Automatisation of Animal Health Surveillance Through Natural Language Processing
Abstract
1 Introduction
1.1 Study Scope: Salmonellosis (S. Dublin) and Pneumonia NOS
2 Materials and Method
2.1 Materials
2.2 Lexical Analysis of APHA 'Surveillance Highlights' Free-Text Datasets
2.3 NLP Tasks: Named-Entity Recognition and Similarity and Relatedness
2.4 Assessing the Performance for the NLP Tasks
3 Results
3.1 Lexical Analysis of APHA Surveillance Highlights Free-Text Datasets
3.2 NLP Tasks: Name-Entity Recognition and Similarity and Relatedness
3.3 Assessing the Performance of the NLP Tasks
4 Discussion
5 Conclusion
Acknowledgements
References
Applications of Evolutionary Algorithms
GenMuse: An Evolutionary Creativity Enhancement Tool
Abstract
1 Introduction
2 The System
3 The Methodologies
3.1 Genetic Algorithms
3.2 Interactive Genetic Algorithms (IGAs)
3.3 Neural Networks
4 Practical Use of the Two Methodologies - GAs and ANNs
4.1 The Genetic Algorithm Used for This Project
4.2 Riff Creation
4.2.1 Encoding Chromosomes
4.2.2 Evaluating the Fitness Value of a Solution
4.3 The Artificial Neural Network
4.3.1 Interpretation of Classification
4.3.2 Training the Artificial Neural Network
5 Results
6 Conclusions
7 Future Work
References
Evolutionary Art with an EEG Fitness Function
Abstract
1 Background
1.1 Context
1.2 Interactive Art Projects
1.3 Related Work
1.4 Genetic Algorithms
1.5 Fitness Function
1.6 Roulette Wheel Selection
1.7 Mutation and Crossover
2 Methodology
2.1 Analysis of Mondrian's Paintings
2.2 Physical Construction
3 Testing
4 Discussion
5 Conclusion
References
A Multi-objective Design of In-Building Distributed Antenna System Using Evolutionary Algorithms
Abstract
1 Introduction
2 Related Work
3 Problem Formulation
4 Optimization Model
4.1 A Solution Representation for IB-DAS
4.2 NSGA II Approach to Solving the Problem
5 Experimental Results
5.1 Pareto-Optimal Front Representation
5.2 Comparison of NSGA II Vs. GA
5.3 Run Time Analysis of NSGA II
6 Conclusion and Future Work
References
Machine Learning for Time Series Data
Investigation of Machine Learning Techniques in Forecasting of Blood Pressure Time Series Data
1 Introduction
2 Forecasting Strategies
3 Machine Learning Algorithms
4 Methodology
4.1 Data Set
4.2 Forecast Model Formulation
4.3 Results
5 Discussion
6 Conclusions
References
Stock Index Forecasting Using Time Series Decomposition-Based and Machine Learning Models
1 Introduction
2 Non-classical Decomposition Models
2.1 Discrete Wavelet Transform
2.2 Empirical Mode Decomposition
2.3 Variational Mode Decomposition
3 Ensemble Forecasting Framework
3.1 Data
3.2 Methodology
4 Results and Discussion
4.1 Performance Comparison of Ensemble Models with Standalone Models
4.2 Comparison of Performance Among Decomposition-Based ANN Models and SVR Models
5 Threats to Validity
6 Conclusions
References
Effective Sub-Sequence-Based Dynamic Time Warping
1 Introduction
2 Background
3 Previous Work
4 Sub-Sequence-Based DTW
5 Time Complexity
6 Evaluation
6.1 Data Sets
6.2 Evaluation Results
7 Conclusion
References
Applications of Machine Learning
Developing a Catalogue of Explainability Methods to Support Expert and Non-expert Users
1 Introduction
2 Related Work
3 Use-Case - Explaining Engineering Notes to Desk-Based Agents
3.1 The Dataset
3.2 Evaluation of Classification Engine
3.3 Results and Discussion
3.4 Co-creation of Explanation Methods
4 The Explanation Engine
4.1 Low-Level Explanations
4.2 High-Level Explanations
4.3 Co-creation of Explanations
5 The Application
6 Evaluation of Explanation Framework
6.1 Results and Discussion
7 Conclusions
References
A Generic Model for End State Prediction of Business Processes Towards Target Compliance
1 Introduction
2 A Framework of End State Prediction of Business Process Data
3 Dataset Description
3.1 Pre-processing
3.2 Feature Extraction
4 Results and Discussion
5 Conclusion
References
An Investigation of the Accuracy of Real Time Speech Emotion Recognition
Abstract
1 Introduction
2 Proposed System
2.1 Hardware and Programming Language
2.2 Voice Detector
2.3 Pre-processing
2.4 Feature Extraction
2.5 Feature Normalisation
2.6 Pattern Recognition Algorithms
3 Experiments
3.1 The Dataset
3.2 Experiments on Speech Databases
3.3 Real Time Speech Emotion Case-Based Study
4 Evaluation
4.1 Results Evaluation
4.2 Performance on the Raspberry Pi 3 B+
4.3 Comparison with Other Speech Emotion Systems
5 Conclusions and Future Work
References
Contributing Features-Based Schemes for Software Defect Prediction
Abstract
1 Introduction
2 Related Work
3 Contributing Features-Based Logistic Regression
3.1 A Case Study of Contributing Feature Selection
4 The Prediction Models
5 Experiments and Analysis
5.1 Results and Discussions
5.2 Comparative Analysis
6 Conclusion
Acknowledgement
References
Induction Motor Inter-turn Short Circuit Fault Detection Using Efficient Feature Extraction for Machine Learning Based Fault Detectors
Abstract
1 Introduction
2 Literature Survey
3 Features Extracted from the GOTIX Vibration Dataset
4 Features Extracted from the GOTIX Electric Dataset
5 Features from the Envelope Spectrums of the Vibration Signals
6 Features from the Average Spectrum Versus Order for Vibration Signals
7 Conclusion
References
Hybrid Feature Selection Method for Improving File Fragment Classification
1 Introduction
2 Related Work
3 Proposed Hybrid Feature Selection Method
3.1 Data Collection/Extraction
3.2 File Fragmentation
3.3 Feature Extraction
3.4 Features Selection
3.5 Classification
4 Experiments and Results
4.1 Data and Feature Extraction
4.2 Experiments for Selecting Appropriate Filters and Classifiers
4.3 Implementing Hybrid Feature Selection Method
4.4 File Fragment Classification Results
5 Conclusion
References
Optimization of Silicon Tandem Solar Cells Using Artificial Neural Networks
Abstract
1 Introduction
2 Previous Works
3 Dataset Description
4 The Architecture of an Artificial Neural Network
5 Results and Discussions
6 Conclusion
References
Stochastic Local Search Based Feature Selection for Intrusion Detection
Abstract
1 Introduction
2 Background
2.1 Classification
2.2 Feature Selection
2.3 Stochastic Local Search
3 Proposed Approach
3.1 Proposed Architecture for Intrusion Detection
3.2 Solution Representation
3.3 SLS for Feature Selection
3.4 SLS with Classifier for Intrusion Detection
3.5 The Considered Classifiers
4 Experiments
4.1 The Considered Datasets
4.2 Evaluation Measures
4.3 Parameters Tuning and Impacts
4.4 Numerical Results
4.5 Further Comparison
5 Conclusion
References
Knowledge Acquisition
Improving the Adaptation Process for a New Smart Home User
1 Introduction
2 Related Work
3 User-Guided Transfer Learning (UTL)
3.1 Survey
3.2 Simulation
3.3 Activity Recognition
3.4 Transfer Learning
4 User-Guided Transfer Learning (UTL) Interface
5 Validation
6 Discussion
7 Conclusion
References
Short Application Papers
Analysis of Electronic Health Records to Identify the Patient's Treatment Lines: Challenges and Opportunities
Abstract
1 Introduction
2 Challenge for Reconstructing the Patient's Treatment Lines
2.1 Named Entity Recognition Challenges
2.2 Temporal Relation Identification Challenges
2.3 The Integration of Structured Results Challenges
3 Solution
4 Conclusion and Future Work
Acknowledgment
References
Characterisation of VBM Algorithms for Processing of Medical MRI Images
1 Introduction
1.1 Image Registration (Co-registration) Algorithms
2 Methodology
2.1 Temporal Cortex
3 Discussion
References
Analogical News Angles from Text Similarity
1 Introduction
2 Assumptions
3 Finding Unmatched News Reports
4 Similarity Measures
5 Results
6 Conclusion and Further Work
References
Mindfulness Mirror
Abstract
1 Introduction
2 Genetic Algorithms
3 Interactive Genetic Algorithms
4 Mindfulness
5 Bio Feedback and Human Issues
6 Construction of the System
7 Results and Discussion
8 Conclusion
References
Predicting Bid Success with a Case-Based Reasoning Approach: Explainability and Business Acceptance
Abstract
1 Background
2 Case-Based Reasoning: An Alternative
3 Case-Based Reasoning Approach
4 User Engagement and Acceptance
5 Conclusions
References
Data Augmentation for Ambulatory EEG Based Cognitive State Taxonomy System with RNN-LSTM
1 Introduction
2 Proposed Work
2.1 Data Collection Procedure
2.2 Proposed Methodology
2.3 Data Augmentation Method
3 Result Analysis and Discussion
4 Conclusion and Future Work
References
Time-Series-Based Classification of Financial Forecasting Discrepancies
1 Introduction
2 Background
3 Dataset Description
4 Methods
5 Experiments
6 Conclusion
References
Predicting Soil pH by Using Nearest Fields
1 Introduction
2 Related Work
3 Predicting Soil pH
3.1 Soil Dataset
3.2 Features Based on Nearest Fields
3.3 Data Mining Techniques for Prediction
4 Experimental Results
5 Conclusion and Future Work
References
Information Retrieval for Evidence-Based Policy Making Applied to Lifelong Learning
1 Introduction
2 The Policy Data Model
2.1 Descriptive Feature Model
2.2 Policy Retrieval Model
3 Query Models for Policy Search
3.1 Free-Text Queries
3.2 Free-Text Structured Queries
3.3 Constrained Structured Queries
4 Evaluation
5 The ENLIVEN Dataset
6 Background and Related Works
7 Conclusion and Research Directions
References
On Selection of Optimal Classifiers
Abstract
1 Introduction
2 Research Background
2.1 Machine Learning
2.2 Pareto Optimality
3 Research Methodology
4 Experiments and Results
4.1 The Dataset
4.2 Results
5 Conclusion and Future Work
References
Author Index

Systemvoraussetzungen

Als PDF speichern Als Link merken