Modeling Decisions for Artificial Intelligence : 18th International Conference, MDAI 2021, Umeå, Sweden, September 27–30, 2021, Proceedings

This book constitutes the refereed proceedings of the 18th International Conference on Modeling Decisions for Artificial Intelligence, MDAI 2021, held in Umeå, Sweden, in September 2021.* The 24 papers presented in this volume were carefully reviewed and selected from 50 submissions. Additionally, 3 invited papers were included. The papers discuss different facets of decision processes in a broad sense and present research in data science, data privacy, aggregation functions, human decision making, graphs and social networks, and recommendation and search. The papers are organized in the following topical sections: aggregation operators and decision making; approximate reasoning; machine learning; data science and data privacy. *The conference was held virtually due to the COVID-19 pandemic. Preface Organization Contents Invited Papers Andness-Directed Iterative OWA Aggregators 1 Introduction 2 The Soft Iterative SItOWA Aggregator and a Recursive Method for Computing Its Weights 3 Andness-Directed SItOWA and HItOWA Aggregators 4 Weighted ItOWA Aggregators 5 Conclusions References New Eliahou Semigroups and Verification of the Wilf Conjecture for Genus up to 65 1 Introduction 2 The Bitstream of Gaps and the Bitstream of Seeds of a Numerical Semigroup 3 A Graphical Explanation of the Algorithm by an Example 4 Eliahou Semigroups and Wilf Conjecture Verification Extended up to Genus 65 References Are Sequential Patterns Shareable? Ensuring Individuals' Privacy 1 Introduction 2 Related Work 3 Background 3.1 Pattern Mining 3.2 Differential Privacy and Noise Graph Mechanism 3.3 Disclosure Risk 3.4 Information Loss and Utility 4 Methodology 5 Experiments and Results 6 Conclusion and Future Work References Aggregation Operators and Decision Making On Two Generalizations for k-Additivity 1 Introduction 2 Definitions and Notations 3 Formulaic k-Additivity of a Constructively k-Additive Set Function 4 k-Additivity of Distorted Measure 5 k-Additivity in a General Case 6 Conclusion References Sequential Decision-Making Under Uncertainty Using Hybrid Probability-Possibility Functions 1 Introduction 2 Decision Trees 3 Hybrid Possibility-Probability Measures 4 Decision Making on Hybrid -p Decision Trees 4.1 Hybrid Utility Functional and Decision Maker Behavior 4.2 Decision Trees with Hybrid -p Distributions 4.3 Composing Possibilistic and Probabilistic Lotteries 5 Conclusion References Numerical Comparison of Idempotent Andness-Directed Aggregators 1 Introduction 2 Imprecision and Uncertainty of Human Percepts 3 The Worst Case Analysis: Soft Idempotent Disjunctive Aggregators 4 Experiments with Strictly Soft Idempotent Disjunctive Aggregators 5 Conclusions References Approximate Reasoning Multiple Testing of Conditional Independence Hypotheses Using Information-Theoretic Approach 1 Introduction 2 Preliminaries 2.1 Conditional Mutual Information 2.2 Multiple Conditional Independence Testing and JMI Statistic 3 Main Result: Dichotomous Behaviour of Test Statistic Statistic JMI"0362JMI 4 Asymptotic Versus Generic Methods 4.1 Asymptotic Method 4.2 Generic Methods 5 Simulation Study 5.1 Artificial Data Sets 5.2 Medical Data Set Example 5.3 Conclusion References A Bayesian Interpretation of the Monty Hall Problem with Epistemic Uncertainty 1 Introduction 2 Epistemic Uncertainty over Monty's Protocol 3 Different Priors for Doors 4 General Setting: n Doors and General Response Model 5 Discussion and Conclusions References How the F-Transform Can Be Defined for Hesitant, Soft or Intuitionistic Fuzzy Sets? 1 Introduction 2 Preliminaries and Categorical Tools 3 F-Transform for Hesitant, Soft and Intuitionistic L-Fuzzy Sets 3.1 Hesitant F-Transform 3.2 Intuitionistic F-Transform 3.3 Soft F-Transform References Enhancing Social Recommenders with Implicit Preferences and Fuzzy Confidence Functions 1 Introduction 2 Methodology 2.1 Data Recollection and Experimental Setting 2.2 Social-Enhanced KNN-CF Algorithms 3 Fuzzy Confidence Functions 4 Results 4.1 Explicit Preferences 4.2 Implicit Preferences 5 Final Remarks References A Necessity Measure of Fuzzy Inclusion Relation in Linear Programming Problems 1 Introduction 2 Preliminaries 2.1 Fuzzy LPP 2.2 Necessity Measure 3 Model Conversion 3.1 Fuzzy LPP by Inclusion Relation 3.2 Conversion Principles 4 Algorithm for the Fuzzy LP 4.1 Special Case with Constant A 4.2 General Case 5 Numerical Example 6 Conclusion References Machine Learning Mass-Based Similarity Weighted k-Neighbor for Class Imbalance 1 Introduction 2 Mass-Based Dissimilarity Measurement 2.1 Definition 2.2 k-lowest masse neighbors (k-LMN) 3 Mass-Based Similarity Weighted k-neighbor (Sk-LMN) 3.1 Confidence Estimation 3.2 Mass-Based Similarity Measurement 3.3 Weighted Sum Aggregation 3.4 Decision Making 4 Experiential Results 4.1 Dataset Description 4.2 Implementation Details and Evaluation Metrics 4.3 Results and Discussions 4.4 Non-parametric Statistical Analysis 5 Conclusion References Multinomial-Based Decision Synthesis of ML Classification Outputs 1 Introduction 2 Multinomial-Based Decision Synthesis for ML Classification Problems 2.1 Classification Decision Framework 2.2 Simplifying Assumptions and Operational Dependencies 2.3 Aggregating Classification Decisions 3 Decision Aggregation of Feed-Forward CNN Classifiers 3.1 Toy Example 3.2 Binomial Reduction Applied to Mask Recognition Algorithm 3.3 Specific Emitter Identification and Modulation Classification 4 Conclusions References Quantile Encoder: Tackling High Cardinality Categorical Features in Regression Problems 1 Introduction 2 Quantile Encoder 2.1 Summary Encoder 3 Experiments 3.1 Dataset Bench-Marking 3.2 Code and Reproducibility 3.3 Comparison of all Encoding Methods 3.4 Encoding Dependence with Respect to the Evaluation Metric 3.5 Summary Encoder Performance 3.6 Discussion 4 Conclusion References Evidential Undersampling Approach for Imbalanced Datasets with Class-Overlapping and Noise 1 Introduction 2 Evidence Theory 3 Evidential Undersampling Approach (EVUS) 3.1 Determination of Centers 3.2 Computing the Soft Labels 3.3 Selecting Majority Samples for Elimination 4 Experimental Study 4.1 Setup 4.2 Results and Discussion 5 Conclusions References Well-Calibrated and Sharp Interpretable Multi-Class Models 1 Introduction 2 Background 2.1 Probabilistic Prediction 2.2 Probability Estimation Trees 2.3 Venn-Abers Predictors 3 Method 4 Results 5 Concluding Remarks References Automated Attribute Weighting Fuzzy k-Centers Algorithm for Categorical Data Clustering 1 Introduction 2 Related Works 3 Preliminaries 3.1 Fuzzy Clustering for Categorical Data 3.2 k-means Algorithm for Fuzzy Clustering (Fk-means) 3.3 Previous Work (Fk-centers Algorithm) 3.4 Attribute Weighting for Clustering 4 The Proposed Method 4.1 Objective Function 4.2 WCFk-centers Algorithm 5 Experiments and Results 5.1 Datasets and Testing Environment 5.2 Evaluation Metrics 5.3 Attribute Weighting Performance 5.4 Fuzzy Clustering Performance Analysis 5.5 Complexity Analysis 6 Conclusion and Future Work References q-Divergence Regularization of Bezdek-Type Fuzzy Clustering for Categorical Multivariate Data 1 Introduction 2 Preliminaries 2.1 Divergence 2.2 Fuzzy Clustering for Vectorial Data 2.3 Conventional Fuzzy Clustering Method for Categorical Multivariate Data 3 Proposed Method 3.1 Basic Concepts 3.2 Algorithm 4 Numerical Experiment 5 Summary References Automatic Clustering of CT Scans of COVID-19 Patients Based on Deep Learning 1 Introduction 2 Related Work 3 Materials 4 Methods 4.1 Feature Extraction with VGG16 4.2 Deep Embedded Clustering 4.3 Visualization via PCA 5 Experiment 5.1 Setting 5.2 Results 6 Conclusion References Network Clustering with Controlled Node Size 1 Introduction 2 Preliminaries 2.1 k-medoids Clustering 2.2 Controlled-Sized Clustering Based on Optimization 2.3 Kernel Method 3 Proposed Method 4 Numerical Experiments 4.1 Experimental Setup 4.2 Experimental Results 4.3 Discussions 5 Conclusions References Data Science and Data Privacy Fairly Private Through Group Tagging and Relation Impact 1 Introduction 2 Methodology 2.1 One-Hot Encoding 2.2 Query Function 2.3 Correlation Matrix 2.4 Group Tagging Method 2.5 Fairly Iterative Shuffling (FIS) 2.6 2-Layers Fair Classification Through Relation Impact 2.7 Risk Minimization and Count Report Generation 3 Analysis 3.1 FIS Randomised Response Ratio and Privacy Budget 3.2 Fair Classifiers and Relation Impact 3.3 Boosting Utility Through Risk Minimization 4 Conclusion and Future Scope References MEDICI: A Simple to Use Synthetic Social Network Data Generator 1 Introduction 2 Background to Third Party Algorithms 2.1 R-MAT 2.2 Louvain 3 Medici 3.1 Overall Vision 3.2 Data Assignment and Propagation 4 Empirical Testing 4.1 Setup, Computational Cost and Performance 4.2 Example Data Generated 4.3 Stability of the Data Generator: Deviation Between Successive Executions 5 Conclusions References Answer Passage Ranking Enhancement Using Shallow Linguistic Features 1 Introduction 2 Methods 2.1 Data Set 2.2 Deep Neural Network Ranker 2.3 Explicit Shallow Linguistic Features 2.4 Fusion of Linguistic Features and Deep Semantics 3 Discussion 4 Conclusions References Neural Embedded Dirichlet Processes for Topic Modeling 1 Introduction 2 Related Work 3 Background 4 Our Models 4.1 The Embedded Dirichlet Process 4.2 The Embedded Hierarchical Dirichlet Process 5 Experiments 5.1 Datasets 5.2 Training Settings 5.3 Metrics 5.4 Results 6 Conclusion and Future Work References Density-Based Evaluation Metrics in Unsupervised Anomaly Detection Contexts 1 Introduction 2 Background Knowledge 3 Evaluation Metrics 4 Experimental Analysis 4.1 Approaches Hyper-Parameters Values Definition 4.2 Datasets: Harvard Dataverse and Numenta Anomaly Benchmark 4.3 Experimental Setup 4.4 Empirical Evaluation 5 Conclusions References Explaining Image Misclassification in Deep Learning via Adversarial Examples 1 Introduction 2 Related Work 3 Our Proposals 3.1 Adversarial Examples 3.2 Explaining Model Predictions on the Developer's Side 3.3 Explaining Model Predictions on the User's Side 4 Experimental Results 4.1 Explanations for the Developer 4.2 Explanations for the User 5 Conclusions and Future Research References Towards Machine Learning-Assisted Output Checking for Statistical Disclosure Control 1 Introduction 2 Rewriting Checking Rules for Synthetic Log Generation 3 Generation of Synthetic Training and Test Data 4 Experimental Work 5 Conclusions and Future Research References Author Index