Explainable AI and Other Applications of Fuzzy Techniques

Proceedings of the 2021 Annual Conference of the North American Fuzzy Information Processing Society, NAFIPS 2021
 
 
Springer (Verlag)
  • erschienen am 27. Juli 2021
  • |
  • XII, 506 Seiten
 
E-Book | PDF mit Wasserzeichen-DRM | Systemvoraussetzungen
978-3-030-82099-2 (ISBN)
 

This book focuses on an overview of the AI techniques, their foundations, their applications, and remaining challenges and open problems. Many artificial intelligence (AI) techniques do not explain their recommendations. Providing natural-language explanations for numerical AI recommendations is one of the main challenges of modern AI. To provide such explanations, a natural idea is to use techniques specifically designed to relate numerical recommendations and natural-language descriptions, namely fuzzy techniques.

This book is of interest to practitioners who want to use fuzzy techniques to make AI applications explainable, to researchers who may want to extend the ideas from these papers to new application areas, and to graduate students who are interested in the state-of-the-art of fuzzy techniques and of explainable AI-in short, to anyone who is interested in problems involving fuzziness and AI in general.


1st ed. 2022
  • Englisch
  • Cham
  • |
  • Schweiz
Springer International Publishing
  • 150
  • |
  • 48 s/w Abbildungen, 150 farbige Abbildungen
  • 42,11 MB
978-3-030-82099-2 (9783030820992)
10.1007/978-3-030-82099-2
weitere Ausgaben werden ermittelt
  • Intro
  • Explainability: New Application and New Promise of Fuzzy Techniques
  • Contents
  • A Fuzzy Logic Approach for Spacecraft Landing Site Selection
  • 1 Introduction
  • 2 Image Dataset
  • 3 Methodology
  • 3.1 Image Size Reduction
  • 3.2 Object Grouping
  • 3.3 Object Classification
  • 3.4 Performance Metrics
  • 4 Results
  • 5 Conclusions and Future Work
  • 5.1 Future Work
  • References
  • Takagi-Sugeno Fuzzy Systems with Triangular Membership Functions as Interpretable Neural Networks
  • 1 Introduction
  • 2 Preliminaries
  • 2.1 Fuzzy Systems
  • 2.2 Neural Networks
  • 3 Equivalence Between TS Fuzzy Systems with Triangular Membership And Neural Networks with ReLU Activation
  • 3.1 TS Fuzzy Systems Expressed in Terms of ReLU Functions
  • 3.2 TS Fuzzy Systems as Neural Networks with ReLU Activation
  • 3.3 Neural Networks Expressed in Terms of Takagi-Sugeno Fuzzy Systems
  • 4 Conclusions
  • References
  • A Study on Constrained Interval Arithmetic
  • 1 Introduction
  • 2 Constrained Interval Arithmetic
  • 2.1 Addition in CIA
  • 2.2 Difference in CIA
  • 2.3 Multiplication in CIA
  • 2.4 Division in CIA
  • 2.5 Additive/Multiplicative Inverse in CIA
  • 3 Distributivity, Expression/component-Wise and Optimization
  • 4 Inclusion Isotonicity
  • 5 Final Remarks
  • References
  • Fuzzy Classification of Multi-intent Utterances
  • 1 Introduction
  • 2 Related Work
  • 3 Utterance Level Fuzzy Memberships
  • 3.1 Membership Functions
  • 3.2 Parameter Generation
  • 4 Single Intent to Multi Intent Utterances
  • 5 Fuzzy Membership Aggregation and Defuzzification
  • 6 Experiments
  • 6.1 Setup
  • 6.2 Data
  • 6.3 Training Details
  • 6.4 Results
  • 7 Conclusion
  • References
  • How Much for a Set: General Case of Decision Making Under Set-Valued Uncertainty
  • 1 Decision Making Under Set-Valued Uncertainty: Formulation of the Problem
  • 2 What Is Known: Cases of Closed Intervals and Closed Sets
  • 3 Extending the Known Result to General Bounded Sets
  • 4 What if We Do Not Require Additivity?
  • Reference
  • A Deep Fuzzy Semi-supervised Approach to Clustering and Fault Diagnosis of Partially Labeled Semiconductor Manufacturing Data
  • 1 Introduction
  • 2 Methodology
  • 2.1 Deep Convolutional Unsupervised Feature Learning
  • 2.2 PCA-Based Semi-supervised Fault Classification
  • 2.3 Fuzzy c-means Clustering and Borderline Case Detection
  • 3 Results
  • 3.1 Case Study Description
  • 3.2 Semi-supervised Classification Results
  • 3.3 Fuzzy c-means Clustering Results
  • 4 Discussion
  • 5 Conclusion
  • References
  • Why Fuzzy Techniques in Explainable AI? Which Fuzzy Techniques in Explainable AI?
  • 1 Why Fuzzy Techniques in Explainable AI
  • 2 Which Fuzzy Techniques in Explainable AI
  • References
  • Why Are Fuzzy and Stochastic Calculus Different?
  • 1 Introduction
  • 2 Introduction to Stochastic Equations and First Issues with Derivatives
  • 3 Modes of Convergence
  • 4 Convergence in Distribution
  • 5 Is Weakening the Topology a Viable Solution?
  • 6 Comparison with Fuzzy Calculus. Common Problems and Conclusion
  • References
  • A Genetic Fuzzy Approach for the Prediction of Heart Failure
  • 1 Introduction
  • 2 Literature Review
  • 3 Methodology
  • 4 Results
  • 5 Conclusion
  • References
  • COVID-19 Vaccination Priority Evaluation
  • 1 Introduction
  • 2 The Vaccination Priority Degree and Its Use
  • 3 The Vaccination Priority Attribute Tree
  • 4 Vaccination Priority Attribute Criteria
  • 5 Logic Aggregation
  • 6 Examples and Discussion
  • 7 Conclusions
  • References
  • Comparing the Explainability and Performance of Reinforcement Learning and Genetic Fuzzy Systems for Safe Satellite Docking
  • 1 Introduction
  • 2 Background
  • 3 Methodology
  • 3.1 Run Time Assurance
  • 3.2 Reinforcement Learning Approach
  • 3.3 Genetic Fuzzy System Approach
  • 3.4 Run Time Assurance Approach
  • 3.5 Reward Function
  • 4 Results
  • 4.1 Training
  • 4.2 Explainability
  • 5 Conclusion
  • References
  • A Note on the KM Algorithm for Computing the Variance of an Interval Type-2 Fuzzy Set
  • 1 Introduction and Motivation
  • 2 Basics of Interval Type-2 Fuzzy Numbers
  • 3 The KM Algorithm
  • 3.1 a) KM Algorithm for Computing yl
  • 3.2 b) KM Algorithm for Computing yr
  • 4 Computing the Variance of F2(X)
  • 5 Illustrative Examples
  • 5.1 Trapezoidal IT2FS
  • 5.2 Nonlinear IT2FS
  • 6 Concluding Remarks
  • References
  • Estimating Fuzzy Possibility Functions for E-Commerce Decision Support
  • 1 Introduction
  • 2 Methodology
  • 2.1 Data Experiments
  • 2.2 Linguistic Pre-processing
  • 2.3 Feature Extraction and Statistical Methods
  • 2.4 Model Evaluation
  • 3 Soft Business Function
  • 4 Results
  • 5 Final Remarks
  • References
  • Machine Learning to Augment the Fusion Process for Data Classification
  • 1 Introduction
  • 2 Dempster-Shafer Fusion
  • 3 Q-Learning
  • 3.1 Q-Learning Fundamentals
  • 3.2 Dempster-Shafer Fusion Calculus Integrated into Q-Learning
  • 4 Experiments
  • 5 Conclusions and Future Research
  • References
  • Managing Uncertainty in Crowdsourcing with Interval-Valued Labels
  • 1 Introduction
  • 1.1 Uncertainties in Crowdsourcing
  • 1.2 Previous Results and Motivations of This Study
  • 2 Interval-Valued Labels and Their Properties
  • 2.1 An Interval-Valued Label and Its Properties
  • 2.2 Statistic and Probabilistic Properties of a Set of IVLs on the Same xi
  • 3 Methods for Pre-processing IVLs and Inference Making
  • 3.1 Strategies on Pre-processing Li
  • 3.2 Making an Inference from Li
  • 4 Computational Experiments
  • 4.1 Comparisons of Inferences with MV and Algorithm2 from the Same Li
  • 4.2 Impacts of Biased IVLs on the Distribution of Matching Probability
  • 5 Summary
  • References
  • Modeling Fuzzy Cluster Transitions for Topic Tracing
  • 1 Introduction
  • 2 Related Work
  • 3 Modeling Cluster Transition with Fuzzy Logic
  • 3.1 Crisp Cluster Transitions for Topic Tracing
  • 3.2 Fuzzified Cluster Transitions
  • 4 Implementation: Comparing Crisp and Fuzzy Cluster Transitions
  • 5 Conclusion
  • References
  • A Natural Formalization of Changing-One's-Mind Leads to Square Root of ``Not'' and to Complex-Valued Fuzzy Logic
  • 1 Formulation of the Problem
  • 2 Analysis of the Problem and the Resulting Derivation of Square Root of ``Not" and of Complex-Valued Fuzzy Logic
  • References
  • Each Realistic Continuous Functional Dependence Implies a Relation Between Some Variables: A Theoretical Explanation of a Fuzzy-Related Empirical Phenomenon
  • 1 Formulation of the Problem
  • 2 Formalization of the Problem
  • 3 Main Result
  • 4 Auxiliary Result: Case of m=1
  • References
  • Uniform Mixture Convergence of Continuously Transformed Fuzzy Systems
  • 1 Rule-Based Probability Mixtures and XAI
  • 2 Probability Structure of Additive Fuzzy Rule-Based Systems
  • 3 Uniform Convergence of Mixtures of Transformed Fuzzy Systems
  • 4 Conclusions
  • References
  • Solutions of Systems of Linear Fuzzy Differential Equations for a Special Class of Fuzzy Processes
  • 1 Introduction
  • 2 Mathematical Background
  • 3 Calculus Theory for Fuzzy Functions Under -derivative
  • 4 Systems of Linear FDEs for S-linearly Correlated Fuzzy Processes
  • 5 Application on Mass-Spring Multiple System
  • 6 Final Remarks
  • References
  • Abstractive Representation Modeling for Image Classification
  • 1 Introduction
  • 1.1 Background
  • 1.2 Identified Research
  • 2 Data
  • 2.1 MNIST Digit Dataset
  • 3 Method
  • 3.1 Abstractive Processing
  • 3.2 Abstractive Level Design
  • 3.3 Pattern Frequency Detector
  • 3.4 Classifier
  • 4 Results
  • 4.1 Global Performance Analysis of ARM and CNN
  • 4.2 Robustness Analysis of ARM and CNN
  • 5 Conclusion
  • 5.1 Contributions
  • 5.2 Future Research
  • References
  • A-Cross Product for Autocorrelated Fuzzy Processes: The Hutchinson Equation
  • 1 Introduction
  • 2 Fuzzy Set Theory
  • 3 A-cross Product of Linearly Correlated Fuzzy Numbers
  • 4 Fuzzy Differential Equations Under Fréchet Derivative
  • 5 The Hutchinson Equation
  • 5.1 Crisp Logistic Growth Rate
  • 5.2 Fuzzy Logistic Growth Rate
  • 6 Final Considerations
  • References
  • Use of T-Norm in an Epidemiological Model for COVID-19
  • 1 Introduction
  • 1.1 Triangular Norm and SIR Model
  • 2 Methodology
  • 3 Application and Results
  • 4 Comments and Discussions
  • 5 Conclusion
  • References
  • Fuzzy-Flatness Hybrid Fault-Tolerant Control
  • 1 Introduction
  • 2 Differential Flatness
  • 3 3-Tank Problem
  • 4 Fuzzy-Flatness Hybrid Control
  • 5 Results
  • 6 Conclusion
  • References
  • Finding Fuzziness in Neural Network Models of Language Processing
  • 1 Introduction
  • 2 The Concept of Vagueness and the Vagueness of Concepts
  • 2.1 Psychological Significance
  • 2.2 Handling Vagueness Using Fuzzy Interpretations of Linguistic Hedges
  • 3 Pre-trained Language Models
  • 4 Vagueness in Natural Language Inference Models
  • 4.1 Methodology
  • 5 General Discussion and Conclusion
  • References
  • Interval Arithmetic: WSM, CI or RDM?
  • 1 Introduction
  • 2 CIA and RDM Arithmetics
  • 2.1 Fuzzy Interval Linear Systems
  • 3 Conclusion
  • References
  • Breast Cancer Wisconsin (Diagnostic) Data Analysis Using GFS-TSK
  • 1 Introduction
  • 2 Literature Review
  • 3 Methodology
  • 4 Results
  • 5 Conclusions
  • References
  • A New GA-PSO Optimization Methodology with Fuzzy Adaptive Inertial Weight
  • 1 Introduction
  • 2 Proposed Optimization Methodology
  • 2.1 Parents Selection Genetic Operator
  • 2.2 Crossover Genetic Operator
  • 2.3 Mutation Genetic Operator
  • 2.4 Velocity and Position Vector Update
  • 2.5 Fuzzy Adaptive Inertial Weight
  • 3 Computational Results
  • 4 Conclusion
  • References
  • Predicting Diabetes Diagnosis with Binary-To-Fuzzy Extrapolations and Weights Tuned via Genetic Algorithm
  • 1 Introduction
  • 1.1 Diabetes Prevalence, Diagnosis, Physical, and Monetary Costs
  • 1.2 Author Exposure
  • 1.3 Disease Prediction Using Intelligent Systems
  • 1.4 Kernels and Fuzzy Equivalence Relations in Fuzzy Logic
  • 2 Diabetes Dataset
  • 2.1 Data Acquisition and Prior Method Comparison
  • 2.2 Clinical Characteristics of Study Subjects
  • 3 Methodology
  • 3.1 Fuzzy Relation Kernels, Optimization, and Prediction
  • 3.2 Genetic Algorithm Parameters
  • 3.3 Additional Metrics
  • 4 Results
  • 4.1 Accuracy of Predictions Using Best Chromosome
  • 4.2 Confusion Matrix and Performance
  • 4.3 Method Comparison
  • 5 Discussion
  • 6 Conclusion
  • References
  • Genetic Fuzzy Hand Gesture Classifier
  • 1 Introduction
  • 2 Literature Review
  • 3 Hand Gesture Data Set
  • 4 Methodology
  • 5 Results
  • 5.1 Training
  • 5.2 Testing
  • 5.3 Data Visualization
  • 6 Conclusion and Future Work
  • References
  • Toward Explainable AI-Genetic Fuzzy Systems-A Use Case
  • 1 Introduction
  • 1.1 Fuzzy Logic
  • 1.2 Genetic Algorithm
  • 1.3 Aggregate Fuzzy Tree
  • 2 Methodology
  • 2.1 The Breast Cancer Data Set
  • 3 Results
  • 3.1 Test Confusion Matrices
  • 3.2 Train Confusion Matrices
  • 3.3 Analysis of Results: Fixed Input Variable Mutation Method
  • 4 Discussion
  • 5 Summary and Conclusions
  • References
  • Uncertainty to Avoid Entrapment: Comparing Internet Stings to Real Victim Conversations
  • 1 Introduction
  • 2 Background
  • 3 Related Work
  • 4 Methodology
  • 4.1 Corpus Composition
  • 4.2 Annotation
  • 5 Preliminary Results
  • 5.1 Friendship Forming
  • 5.2 Relationship Forming
  • 5.3 Exclusivity
  • 5.4 Sexual
  • 5.5 Meeting
  • 5.6 Risk Assessment
  • 6 Conclusion and Future Work
  • References
  • FLTRL: A Fuzzy-Logic Transfer Learning Powered Reinforcement Learning Method for Intelligent Online Control in Power Systems
  • 1 Introduction
  • 2 The Problem Settings
  • 3 The Reinforcement-Learning-Based Control Mechanism
  • 4 Fuzzy-Logic-Based Transfer Learning Mechanism
  • 4.1 Transfer Learning Model
  • 4.2 Fuzzy Strategy
  • 5 Simulation Results
  • 5.1 Demand Surge
  • 5.2 Generator Fault Alongside Demand Surge
  • 6 Conclusions
  • References
  • Genetic Fuzzy Methodology to Predict Time to Return to Play from Sports-Related Concussion
  • 1 Introduction
  • 2 Methodology
  • 2.1 Dataset Description
  • 2.2 Data Reduction
  • 2.3 Fuzzy Bolt©
  • 2.4 Regression Problem: Predicting the RTP Days
  • 2.5 Classification Problem
  • 3 Results
  • 3.1 Evaluating the Trained GFS on the Regression Problem
  • 3.2 Evaluating the Trained GFS on the Classification Problem
  • 4 Conclusions
  • References
  • Fuzzy Information Processing Computing Curricula: A Perspective from the First Two-Years in Computing Education
  • 1 Introduction
  • 2 Challenges and Opportunities
  • 3 Computer Science Practices: Learning Outcomes, Alignment, and Adaptation
  • 4 Discussion, Current Efforts, and Future Work
  • References
  • What Teachers Can Learn from Machine Learning
  • 1 Introduction
  • 2 First Idea: Take into Account that There Are Very Wrong and Somewhat Wrong Answers
  • 3 Second Idea: Asking Why-Questions, Not Just Checking Where Answers Are Correct
  • 4 Third Idea: Let Us Be Positive
  • 5 Fourth Idea: Making Learning More Robust
  • 6 Fifth Idea: Averaging - This Is Not What You Think
  • 7 What Else?
  • References
  • Self-tuned Model-Based Predictive Control Using Evolving Fuzzy Model of a Non-linear Dynamic Process
  • 1 Introduction
  • 2 Evolving Fuzzy Model Identification
  • 2.1 ARX Linear Model
  • 2.2 RLS
  • 2.3 Filtering of Regressors
  • 2.4 Non-linear Model
  • 2.5 Model Excitation for Identification
  • 2.6 Model Merging
  • 2.7 Model Combination
  • 3 Predictive Functional Controller
  • 3.1 PFC Derivation
  • 3.2 PFC Derivation for the Evolved Fuzzy Model
  • 4 Evaluation
  • 4.1 Wienner-Hammerstein Dynamic Process
  • 4.2 Evolving Fuzzy Model Identification
  • 4.3 Reference Tracking Using PFC
  • 4.4 Disturbance Rejection Performance
  • 5 Conclusions
  • References
  • A Method to Optimize and Automate the Distribution of Radiology Studies
  • 1 Introduction
  • 2 Technologies
  • 3 Prototype System
  • 4 Data
  • 5 Simulation
  • 6 Conclusions
  • References
  • Fuzzy Baselines to Stabilize Policy Gradient Reinforcement Learning
  • 1 Introduction
  • 2 Learning and Reinforcement Frameworks
  • 2.1 Reinforcement Learning
  • 2.2 Imitation Learning
  • 2.3 Fuzzy Reinforcement Learning
  • 3 Fuzzy Baseline Method for Policy Gradient
  • 4 Computational Experiments
  • 5 Conclusion
  • References
  • Fuzzy Tunes
  • 1 Introduction
  • 2 Review
  • 3 Methodology
  • 3.1 Spectral Bandwidth
  • 3.2 Spectral Centroid
  • 3.3 Chroma Shift
  • 3.4 System Overview
  • 4 Results
  • 5 Conclusion
  • References
  • Mexican Folk Arithmetic Algorithm Makes Perfect Sense
  • 1 Formulation of the Problem
  • 2 Let Us Look at How Computers Add Numbers
  • 3 Let Us Look at Practical Situations Where People Use Addition and Subtraction
  • 4 In Many Practical Situations, We Operate Under Uncertainty
  • 5 Conclusion
  • References
  • Fuzzy-Based, Noise-Resilient, Explainable Algorithm for Regression
  • 1 Introduction
  • 2 Research Objectives
  • 3 Methodology
  • 3.1 The Algorithm
  • 3.2 Comparison to Other Methods
  • 4 Results
  • 4.1 Benchmark with Neural Networks
  • 4.2 Training Stability
  • 4.3 Discussion
  • 4.4 Scalability
  • 5 Conclusion
  • References
  • Evaluation Criteria for Noise Resilience in Regression Algorithms
  • 1 Introduction
  • 2 Research Objectives
  • 3 Methodology
  • 3.1 Framework
  • 3.2 Conditions of Noise Resilience
  • 3.3 NR, Global Noise Resilience Score
  • 4 Conclusion
  • References
  • Numerical Solution for Fuzzy Partial Differential Equations with Interactive Fuzzy Boundary Conditions
  • 1 Introduction
  • 2 Mathematical Background
  • 2.1 Finite Difference Method
  • 2.2 Fuzzy Set Theory
  • 3 Numerical Method to Solve Fuzzy Partial Differential Equation
  • 4 Final Remarks
  • References
  • Fuzzy Logic Leads to a More Adequate Way of Processing Likert-Scale Values: Case Study of Burnout
  • 1 Formulation of the Problem
  • 2 Analysis of the Problem
  • 3 Resulting Recommendation
  • References
  • Author Index

Dateiformat: PDF
Kopierschutz: Wasserzeichen-DRM (Digital Rights Management)

Systemvoraussetzungen:

Computer (Windows; MacOS X; Linux): Verwenden Sie zum Lesen die kostenlose Software Adobe Reader, Adobe Digital Editions oder einen anderen PDF-Viewer Ihrer Wahl (siehe E-Book Hilfe).

Tablet/Smartphone (Android; iOS): Installieren Sie die kostenlose App Adobe Digital Editions oder eine andere Lese-App für E-Books (siehe E-Book Hilfe).

E-Book-Reader: Bookeen, Kobo, Pocketbook, Sony, Tolino u.v.a.m. (nur bedingt: Kindle)

Das Dateiformat PDF zeigt auf jeder Hardware eine Buchseite stets identisch an. Daher ist eine PDF auch für ein komplexes Layout geeignet, wie es bei Lehr- und Fachbüchern verwendet wird (Bilder, Tabellen, Spalten, Fußnoten). Bei kleinen Displays von E-Readern oder Smartphones sind PDF leider eher nervig, weil zu viel Scrollen notwendig ist. Mit Wasserzeichen-DRM wird hier ein "weicher" Kopierschutz verwendet. Daher ist technisch zwar alles möglich - sogar eine unzulässige Weitergabe. Aber an sichtbaren und unsichtbaren Stellen wird der Käufer des E-Books als Wasserzeichen hinterlegt, sodass im Falle eines Missbrauchs die Spur zurückverfolgt werden kann.

Weitere Informationen finden Sie in unserer E-Book Hilfe.


Download (sofort verfügbar)

213,99 €
inkl. 7% MwSt.
Download / Einzel-Lizenz
PDF mit Wasserzeichen-DRM
siehe Systemvoraussetzungen
E-Book bestellen