Kooperativer Bibliotheksverbund

UID:

Format: VIII, 429 p. 170 illus., 138 illus. in color. , online resource.

Edition: 1st ed. 2021.

ISBN: 9783030649494

Series Statement: Studies in Computational Intelligence, 937

Content: This book offers a comprehensive treatise on the recent pursuits of Artificial Intelligence (AI) - Explainable Artificial Intelligence (XAI) by casting the crucial features of interpretability and explainability in the original framework of Granular Computing. The innovative perspective established with the aid of information granules provides a high level of human centricity and transparency central to the development of AI constructs. The chapters reflect the breadth of the area and cover recent developments in the methodology, advanced algorithms and applications of XAI to visual analytics, knowledge representation, learning and interpretation. The book appeals to a broad audience including researchers and practitioners interested in gaining exposure to the rapidly growing body of knowledge in AI and intelligent systems.

Note: Visualizing the Behavior of Convolutional Neural Networks for Time Series Forecasting -- Beyond Deep Event Prediction: Deep Event Understanding based on Explainable Artificial Intelligence -- Interpretation of SVM to build an Explainable AI via Granular Computing -- Factual and Counterfactual Explanation of Fuzzy Information Granules -- Survey of Explainable Machine Learning with Visual and Granular Methods beyond Quasi-explanations -- MiBeX: Malware-inserted Benign Datasets for Explainable Machine Learning -- A Generative Model Based Approach for Zero-shot Breast Cancer Segmentation Explaining Pixels' Contribution to the Model's Prediction.

In: Springer Nature eBook

Additional Edition: Printed edition: ISBN 9783030649487

Additional Edition: Printed edition: ISBN 9783030649500

Additional Edition: Printed edition: ISBN 9783030649517

Language: English

Subjects: Computer Science

Keywords: Aufsatzsammlung

DOI: 10.1007/978-3-030-64949-4

URL: https://doi.org/10.1007/978-3-030-64949-4

URL: Volltext  (URL des Erstveröffentlichers)

UID:

Format: viii, 429 Seiten : , Illustrationen, Diagramme (überwiegend farbig).

ISBN: 978-3-030-64948-7

Series Statement: Studies in computational intelligence volume 937

Additional Edition: Erscheint auch als Online-Ausgabe ISBN 978-3-030-64949-4

Language: English

Subjects: Computer Science

Keywords: Aufsatzsammlung

UID:

Format: 1 online resource (viii, 429 pages) : , illustrations

ISBN: 3-030-64949-0

Series Statement: Studies in Computational Intelligence ; v.937

Note: Intro -- Preface -- Contents -- Explainable Artificial Intelligence for Process Mining: A General Overview and Application of a Novel Local Explanation Approach for Predictive Process Monitoring -- 1 Introduction -- 2 Background and Related Work -- 2.1 Process Mining -- 2.2 Predictive Business Process Management -- 2.3 Deep Learning for Predictive BPM and XAI -- 3 A Framework for Explainable Process Predictions -- 4 A Novel Local Post-Hoc Explanation Method -- 4.1 Binary Classification with Deep Learning -- 4.2 Local Region Identification by Using Neural Codes -- 4.3 Local Surrogate Model -- 5 Experiment Setting -- 5.1 Use Case: Incident Management -- 5.2 Evaluation Measures -- 5.3 Results -- 6 Discussion -- 7 Conclusion -- References -- Use of Visual Analytics (VA) in Explainable Artificial Intelligence (XAI): A Framework of Information Granules -- 1 Introduction -- 2 Explainability Strategies -- 2.1 Feature Selection -- 2.2 Performance Analysis -- 2.3 Model Explanations -- 3 Global and Local Interpretability -- 3.1 Information Scalability -- 3.2 Visual Scalability -- 4 Stability of Explanation -- 5 Visual Analytics for Granular Computing -- 6 Summary -- References -- Visualizing the Behavior of Convolutional Neural Networks for Time Series Forecasting -- 1 Introduction -- 2 Introduction to Neural Networks and Forecasting -- 2.1 Power Time Series Forecasting -- 2.2 Neural Networks -- 3 Relevant Literature -- 4 Training the cnn ae -- 4.1 Experiment -- 4.2 Data and Code -- 4.3 Setup -- 5 Visualization and Patterns -- 5.1 How to Interpret the Visualizations -- 5.2 Input Visualization -- 5.3 Kernel Visualization -- 5.4 Forecast Visualization -- 5.5 Activation Maps -- 5.6 How to Use the Individual Visualizations -- 6 Conclusion -- References -- Beyond Deep Event Prediction: Deep Event Understanding Based on Explainable Artificial Intelligence. , 1 Introduction -- 2 Why Current Machine Learning is Differentiated from Human Learning -- 3 Beyond Deep Event Prediction -- 4 Big Data, AI, and Critical Condition -- 5 DUE Architecture -- 6 Properties of DUE -- 7 The Concept of DUE -- 7.1 Human Critical Thinking -- 7.2 Contextual Understanding -- 8 Learning Model for DUE -- 8.1 Fundamental Computing for DUE -- 8.2 Computing Using CBNs-Based XAI -- 9 DUE Trends and Future Outlooks -- 9.1 Disasters -- 9.2 Economic Consequences -- 9.3 Safety and Security -- 10 Conclusions -- References -- Interpretation of SVM to Build an Explainable AI via Granular Computing -- 1 Introduction -- 1.1 The Era of Explainable AI with Granular Computing -- 2 The Problem with a Gap in Explainability -- 3 Related Work -- 4 Background -- 4.1 SVM Algorithm -- 4.2 Granular Computing -- 4.3 Syllogisms -- 4.4 Explainable Artificial Intelligence -- 5 Research Methodologies -- 5.1 A Constructive Approach in Developing XAI -- 5.2 A Human-Centric Approach at Early Development Stage -- 6 Implementation: A Syllogistic Approach to Interpret SVM's Classification from Information Granules -- 6.1 Data Selection -- 6.2 Identifying the Information Granules from These Data Sets -- 6.3 Analyzing and Interpretation of Syllogisms from SVM -- 6.4 The General Framework for Modelling Syllogistic Rules -- 6.5 Validating the Interpreted Syllogistic Rules with Physicians and CPGs -- 6.6 XAI Knowledge Base for CAD -- 6.7 XAI with Inference Engine -- 6.8 User Interface in Mobile Application -- 6.9 Preliminary Results -- 6.10 Iterative Retuning and Validation of XAI Mobile App with Physicians in the Loop -- 7 Final XAI Mobile App -- 7.1 XAI Mobile App -- 8 Testing Results from XAI Mobile App -- 8.1 Testing Phase I -- 8.2 Testing Phase II -- 8.3 Testing Phase III -- 8.4 Results from Testing -- 9 Conclusion and Discussion -- 10 Future Work -- References. , Factual and Counterfactual Explanation of Fuzzy Information Granules -- 1 Introduction -- 2 Background -- 3 Proposal -- 4 Illustrative Use Case -- 5 Experiments -- 5.1 Experimental Settings -- 5.2 Experiment 1: Relevance of Expert Knowledge-Based Counterfactual Explanations -- 5.3 Experiment 2: An Impact of Posterior Linguistic Approximation -- 6 Discussion -- 7 Conclusion and Future Work -- References -- Transparency and Granularity in the SP Theory of Intelligence and Its Realisation in the SP Computer Model -- 1 Introduction -- 2 Introduction to Transparency -- 3 Introduction to Granularity -- 4 The SP System in Brief -- 4.1 Information Compression -- 4.2 Abstract View of the SP System -- 4.3 Basic Structures in the SP System for Representing Knowledge -- 4.4 The Concept of SP-Multiple-Alignment -- 4.5 Unsupervised Learning -- 4.6 Existing and Potential Strengths of the SP System -- 4.7 SP-Neural -- 4.8 Future Developments -- 5 Information Compression and the Representation and Processing of Knowledge in the SP System -- 5.1 Information Compression via the Matching and Unification of Patterns -- 5.2 Discontinuous Patterns -- 5.3 Seven Variants of ICMUP -- 5.4 The DONSVIC Principle -- 5.5 Ideas Related to the Concept of a Granule -- 5.6 Tying Things Together? -- 6 Transparency via Audit Trails -- 7 Transparency via Granularity and Familiarity -- 7.1 Granularity, Familiarity, and Basic ICMUP -- 7.2 Granularity, Familiarity, and Chunking-With-Codes -- 7.3 Granularity, Familiarity, and Schema-Plus-Correction -- 7.4 Granularity, Familiarity, and Run-Length Encoding -- 7.5 Granularity, Familiarity, and Part-Whole Hierarchies -- 7.6 Granularity, Familiarity, and Class-Inclusion Hierarchies -- 7.7 Granularity, Familiarity, and SP-multiple-alignments -- 8 Interpretability and Explainability -- 9 Conclusion -- References. , Survey of Explainable Machine Learning with Visual and Granular Methods Beyond Quasi-Explanations -- 1 Introduction -- 1.1 What Are Explainable and Explained? -- 1.2 Types of Machine Learning Models -- 1.3 Informal Definitions -- 1.4 Formal Operational Definitions -- 1.5 Interpretability and Granularity -- 2 Foundations of Interpretability -- 2.1 How Interpretable Are the Current Interpretable Models? -- 2.2 Domain Specificity of Interpretations -- 2.3 User Centricity of Interpretations -- 2.4 Types of Interpretable Models -- 2.5 Using Black-Box Models to Explain Black Box Models -- 3 Overview of Visual Interpretability -- 3.1 What is Visual Interpretability? -- 3.2 Visual Versus Non-Visual Methods for Interpretability and Why Visual Thinking -- 3.3 Visual Interpretation Pre-Dates Formal Interpretation -- 4 Visual Discovery of ML Models -- 4.1 Lossy and Lossless Approaches to Visual Discovery in n-D Data -- 4.2 Theoretical Limitations -- 4.3 Examples of Lossy Versus Lossless Approaches for Visual Model Discovery -- 5 General Line Coordinates (GLC) -- 5.1 General Line Coordinates to Convert n-D Points to Graphs -- 5.2 Case Studies -- 6 Visual Methods for Traditional Machine Learning -- 6.1 Visualizing Association Rules: Matrix and Parallel Sets Visualization for Association Rules -- 6.2 Dataflow Tracing in ML Models: Decision Trees -- 6.3 IForest: Interpreting Random Forests via Visual Analytics -- 6.4 TreeExplainer for Tree Based Models -- 7 Traditional Visual Methods for Model Understanding: PCA, t-SNE and Related Point-to-Point Methods -- 8 Interpreting Deep Learning -- 8.1 Understanding Deep Learning via Generalization Analysis -- 8.2 Visual Explanations for DNN -- 8.3 Rule-Based Methods for Deep Learning -- 8.4 Human in the Loop Explanations -- 8.5 Understanding Generative Adversarial Networks (GANs) via Explanations. , 9 Open Problems and Current Research Frontiers -- 9.1 Evaluation and Development of New Visual Methods -- 9.2 Cross Domain Pollination: Physics & -- Domain Based Methods -- 9.3 Cross-Domain Pollination: Heatmap for Non-Image Data -- 9.4 Future Directions -- 10 Conclusion -- References -- MiBeX: Malware-Inserted Benign Datasets for Explainable Machine Learning -- 1 Introduction -- 2 Background and Related Works -- 2.1 Malware Analysis Overview -- 2.2 Granularity in Malware Analysis -- 2.3 Feature Visualization -- 2.4 Malware as Video -- 2.5 MetaSploit -- 2.6 Bash Commands -- 3 Dataset Generation -- 3.1 Gathering Benign Files -- 3.2 Trojan Insertion -- 3.3 Malware Verification -- 3.4 Dataset Generation Results -- 4 Malware Classification -- 4.1 Pre-processing -- 4.2 Network Specifications -- 4.3 Classification Results -- 5 Saliency Mapping -- 6 Conclusion and Future Work -- References -- Designing Explainable Text Classification Pipelines: Insights from IT Ticket Complexity Prediction Case Study -- 1 Introduction -- 2 Related Work -- 2.1 Explainability and Granularity -- 2.2 Text Representation -- 2.3 Text Classification -- 2.4 Ticket Classification Research -- 2.5 Summary -- 3 Methods -- 3.1 Feature Extraction -- 3.2 Machine Learning Classifiers -- 4 Experimental Evaluation -- 4.1 Case Study and Datasets -- 4.2 Experimental Settings -- 4.3 Comparison of SUCCESS and QuickSUCCESS -- 4.4 Results -- 5 Discussion -- 5.1 Explainability and Granularity Implications -- 5.2 Methodological Contributions -- 5.3 Managerial and Practical Contributions -- 6 Conclusion and Future Works -- Appendix I: Taxonomy of Decision-Making Logic Levels -- Appendix II: Business Sentiment Lexicon with Assigned Valences -- References -- A Granular Computing Approach to Provide Transparency of Intelligent Systems for Criminal Investigations -- 1 Introduction. , 2 Supporting Intelligence Analysts with Intelligent Systems.

Additional Edition: ISBN 3-030-64948-2

Language: English

Keywords: Llibres electrònics