Modeling Decisions for Artificial Intelligence : 20th International Conference, MDAI 2023, Umeå, Sweden, June 19–22, 2023, Proceedings

This book constitutes the refereed proceedings of the 20th International Conference on Modeling Decisions for Artificial Intelligence, MDAI 2023, held in Umeå, Sweden, during June19–22,2023. The 17 papers presented in this volume were carefully reviewed and selected from 28 submissions. Additionally, 1 invited paper 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: Decision making and uncertainty; Machine Learning and data science; and Data privacy. Preface Organization Invited Talks Partially Observing Graphs - When Can We Infer Underlying Community Structure? AI for the Public Good - Reflections on Ethics, Decision Making and Work Imprecise Probability in Formal Argumentation Contents Plenary Talk Logic Aggregators and Their Implementations 1 Introduction 2 Geometric Characterization of Logic Aggregators and the Set of Border Aggregators 3 Semantic Identity and Noncommutativity 4 Classification of Graded Logic Functions and Andness-Directedness of Logic Aggregators 5 Specification of Requirements for Logic Aggregators 6 Benchmark Problems for Logic Aggregators 7 Implementation of Logic Aggregators Using GCD and Andness-Directed Interpolative Aggregation 7.1 Andness-Directed Interpolative Aggregation (ADIA) 7.2 Five Conjunctive Border Aggregators 7.3 Four Families of Conjunctive Interpolative Aggregators 7.4 Solutions of Benchmark Problems 7.5 Implementation of Hyperconjunction Using Andness-Directed t-norms 8 Implementations of Logic Aggregators Using OWA Family of Aggregators 8.1 WOWA: OWA with Importance Weights 8.2 Distribution of OWA Weights 8.3 Annihilators for OWA: GOWA and OWG 8.4 Andness-Directed OWA with Annihilators 8.5 The OWA Family for GCD 8.6 Solutions of Benchmark Problems 9 Implementations of Logic Aggregators Using Fuzzy Integrals 10 Implementations of Logic Aggregators Using Means 11 Evaluation and Comparison of Logic Aggregators 12 Aggregation as a Graded Propositional Calculus 13 Conclusions References Decision Making and Uncertainty Multi-target Decision Making Under Conditions of Severe Uncertainty 1 Introduction 2 Decision Making Under Weakly Structured Information 2.1 Weakly Structured Probabilistic Information 2.2 Weakly Structured Preferences 2.3 A Criterion for Decision Making 2.4 Computation 3 Adaptation to Multi-target Decision Making 3.1 Multi-target Decision Making 3.2 Transferring the Concepts 4 Example: Comparison of Algorithms 5 Concluding Remarks References Constructive Set Function and Extraction of a k-dimensional Element 1 Introduction 2 Constructive Set Function 3 Distortion Measures and Constructive Set Functions 4 Extraction of k-dimensional Elements 5 Conclusion References Coherent Upper Conditional Previsions Defined by Fractal Outer Measures to Represent the Unconscious Activity of Human Brain*-12pt 1 Introduction 2 The Model Based on Hausdorff Outer Measures and the Selective Attention 3 Fractal Measures and Dimensions 3.1 Examples and Motivations 3.2 A General Hausdorff Measure, Packing Measure and Dimensions 3.3 General Hewitt-Stromberg Measures and Dimensions 4 Mathematical Representation of Unexpected Events 5 Conclusions References Discrete Chain-Based Choquet-Like Operators 1 Introduction 2 Preliminaries 3 Discrete Chain-Based Choquet-Sugeno-Like Operators 3.1 ChainC-operator of Type 1 3.2 ChainC-operator of Type 2 3.3 Basic Properties of ChainC-operator of Type 1 and 2 3.4 Linearity Property 3.5 (j)ChainC-Operator as Monotone Measure Extension 4 Conclusion References On a New Generalization of Decomposition Integrals 1 Introduction 2 Preliminaries 3 Decomposition Integrals Generalized by Set-Based Extended Aggregation Functions 4 Concluding Remarks References Bipolar OWA Operators with Continuous Input Function 1 Introduction 2 Basic Notations and Some Known Facts 2.1 OWA Operators, Choquet Integral and Bipolar Capacities 2.2 OWA with Continuous Input Functions 2.3 Bipolar Capacity and Bipolar OWA (BIOWA) Operators 2.4 Orness Measures 3 BIOWA with Continuous Input Functions 4 Orness Measure for BIOWA 5 Conclusions References Machine Learning and Data Science Cost-constrained Group Feature Selection Using Information Theory 1 Introduction 2 Related Work 3 Proposed Methods 3.1 Method 1: Single Feature Selection 3.2 Method 2: Group Feature Selection 3.3 Approximating the Relevance Terms 3.4 Algorithms 4 Experiments 4.1 Data 4.2 Results 5 Conclusions References Conformal Prediction for Accuracy Guarantees in Classification with Reject Option 1 Introduction 2 Background 2.1 Probabilistic Prediction and Calibration 2.2 Conformal Classification 2.3 Related Work 3 Method 4 Results 5 Concluding Remarks References Adapting the Gini's Index for Solving Predictive Tasks 1 Introduction 2 Preliminaries 2.1 Notation 2.2 The Lazy Induction of Descriptions Method 2.3 The Gini's Index 3 P-LID: Lazy Induction of Description for Prediction 4 Experiments 5 Conclusions References Bayesian Logistic Model for Positive and Unlabeled Data 1 Introduction 1.1 Notation and Assumptions in PU Learning 1.2 Logistic Model Assumption for PU Data 1.3 Methods of Label Frequency Estimation 2 Gibbs Sampler for Estimation of Label Frequency 2.1 Gibbs Sampling 2.2 Gibbs Sampler for Bayesian Logistic Regression 2.3 Gibbs Sampler for PU Data 3 Numerical Experiments 3.1 Example 3.2 Real Data Simulations 4 Conclusions References A Goal-Oriented Specification Language for Reinforcement Learning 1 Introduction 2 Related Work 3 Reinforcement Learning 4 Case Study - Lunar Lander 5 Environment and Requirements 6 Goal-oriented Specification 6.1 Goals 6.2 Operators 6.3 Annotations 7 Abstraction of Technical Details 8 Conclusion References Improved Spectral Norm Regularization for Neural Networks 1 Introduction 2 Background 2.1 Regularization 3 Method 3.1 Exact Spectral Norm Regularization 3.2 Extension to Non-piecewise Linear Transforms 4 Experiments 4.1 Generalization 4.2 Robustness 5 Conclusion A Experimental details B Conversion Between Operators C Time Efficiency and Relative Error D Proof for Extension Scheme References Preprocessing Matters: Automated Pipeline Selection for Fair Classification 1 Introduction 1.1 Related Work 1.2 Fairness 1.3 Problem Formulation 1.4 Contributions and Overview of Results 2 FairPipes 2.1 The FairPipes Algorithm 3 Experimental Evaluation 3.1 Baseline Mapping of the Search Space 4 Performance Evaluation 5 Conclusions References Predicting Next Whereabouts Using Deep Learning 1 Introduction 2 State of the Art 3 Methodology 4 Attention and Possible Directions for TRAJectory 4.1 Node Self-attention Module 4.2 Possible Directions Module 4.3 Prediction Module 5 Experiments and Results 5.1 Hyperparameter Tuning 5.2 Preliminary Results 6 Conclusions and Future Works References A Generalization of Fuzzy c-Means with Variables Controlling Cluster Size 1 Introduction 2 Preliminaries 3 mGFCM: A Generalization of mSFCM and GFCM, and mRFCM: A Generalization of mEFCM and RFCM 4 Numerical Experiment 5 Conclusion References Data Privacy Local Differential Privacy Protocol for Making Key–Value Data Robust Against Poisoning Attacks 1 Introduction 2 Local Differential Privacy 2.1 Fundamental Definition 2.2 PrivKV 2.3 Poisoning Attack 3 Proposed Algorithm 3.1 Idea 3.2 Oblivious Transfer 3.3 EM Estimation for Key–Value Data 4 Evaluation 4.1 Data 4.2 Methodology 4.3 Experimental Results 4.4 Discussion 5 Conclusion References Differentially Private Graph Publishing Through Noise-Graph Addition*-12pt 1 Introduction and Related Work 2 Basic Definitions 2.1 Noise-Graph Addition 3 Differential Privacy, Sparseness, Random Perturbation and Sparsification 3.1 Sparseness of Randomized Graphs 3.2 Random Perturbation and Random Sparsification 4 Experimental Evaluation 5 Conclusions References Author Index This book constitutes the refereed proceedings of the 9th International Conference on Modeling Decisions for Artificial Intelligence, MDAI 2012, held in Girona, Catalonia, Spain, in November 2012. The 32 revised full papers were carefully reviewed and selected from 49 submissions and are presented with 4 plenary talks. The papers are organized in topical sections on aggregation operators, integrals, data privacy and security, reasoning, applications, and clustering and similarity.