Explainable Artificial Intelligence (XAI) : Concepts, Enabling Tools, Technologies and Applications.

Format:

Book

Author/Creator:

Raj, Pethuru.

Contributor:

Köse, Utku.

Sakthivel, Usha.

Nagarajan, Susila.

Asirvadam, Vijanth Sagayan.

Series:

Computing and Networks Series

Language:

English

Subjects (All):

Artificial intelligence.

Machine learning.

Physical Description:

1 online resource (465 pages)

Edition:

1st ed.

Place of Publication:

Stevenage : Institution of Engineering & Technology, 2023.

Summary:

This book focuses on Explainable AI (XAI) concepts, tools, frameworks, techniques and applications. It introduces knowledge graphs (KG) to support the need for trust and transparency into the functioning of AI systems, and shows how Intelligent applications can be used to greater effect in finance and healthcare.

Contents:

Intro

Title

Copyright

Contents

About the editors

Preface

1 An overview of past and present progressions in XAI

1.1 Introduction

1.2 Background study

1.2.1 Key-related ideas of XAI

1.3 Overview of XAI

1.4 History of XAI

1.5 Top AI patterns

1.6 Conclusion

References

2 Demystifying explainable artificial intelligence (EAI)

2.1 Introduction

2.1.1 An overview of artificial intelligence

2.1.2 Introduction to explainable AI

2.2 Concept of XAI

2.3 Explainable AI (EAI) architecture

2.4 Learning techniques

2.5 Demystifying EAI methods

2.5.1 Clever Hans

2.5.2 Different users and goals in EAI

2.5.3 EAI as quality assurance

2.6 Implementation: how to create explainable solutions

2.6.1 Method taxonomy

2.6.2 Rules - intrinsic local explanations

2.6.3 Prototypes

2.6.4 Learned representation

2.6.5 Partial dependence plot - global post-hoc explanations

2.6.6 Feature attribution (importance)

2.7 Applications

2.8 Conclusion

3 Illustrating the significance of explainable artificial intelligence (XAI)

3.1 Introduction

3.2 The growing power of AI

3.3 The challenges and concerns of AI

3.4 About the need for AI explainability

3.5 The importance of XAI

3.6 The importance of model interpretation

3.6.1 Model transparency

3.6.2 Start with interpretable algorithms

3.6.3 Standard techniques for model interpretation

3.6.4 ROC curve

3.6.5 Focus on feature importance

3.6.6 Partial dependence plots (PDPs)

3.6.7 Global surrogate models

3.6.8 Criteria for ML model interpretation methods

3.7 Briefing feature importance scoring methods

3.8 Local interpretable model-agnostic explanations (LIMEs)

3.9 SHAP explainability algorithm

3.9.1 AI trust with symbolic AI.

3.10 The growing scope of XAI for the oil and gas industry

3.10.1 XAI for the oil and gas industry

3.11 Conclusion

Bibliography

4 Inclusion of XAI in artificial intelligence and deep learning technologies

4.1 Introduction

4.2 What is XAI?

4.3 Why is XAI important?

4.4 How does XAI work?

4.5 Role of XAI in machine learning and deep learning algorithm

4.6 Applications of XAI in machine learning in deep learning

4.7 Difference between XAI and AI

4.8 Challenges in XAI

4.9 Advantages of XAI

4.10 Disadvantages of XAI

4.11 Future scope of XAI

4.12 Conclusion

5 Explainable artificial intelligence: tools, platforms, and new taxonomies

5.1 Introduction

5.2 ML-based systems and awareness

5.3 Challenges of the time

5.3.1 Requirement of explainability

5.3.2 Impact of high-stake decisions

5.3.3 Concerns of society

5.3.4 Regulations and interpretability issue

5.4 State-of-the-art approaches

5.5 Assessment approaches

5.6 Drivers for XAI

5.6.1 Tools and frameworks

5.7 Discussion

5.7.1 For researchers outside of computer science: taxonomies

5.7.2 Taxonomies and reviews focusing on specific aspects

5.7.3 Fresh perspectives on taxonomy

5.7.4 Taxonomy levels at new levels

5.8 Conclusion

6 An overview of AI platforms, frameworks, libraries, and processes

6.1 Introduction to AI

6.2 Role of AI in the 21st century

6.2.1 The 2000s

6.2.2 The 2010s

6.2.3 The future

6.3 How AI transformed the world

6.3.1 Transportation

6.3.2 Finance

6.3.3 Healthcare

6.3.4 Intelligent cities

6.3.5 Security

6.4 AI process

6.5 TensorFlow

6.5.1 Installation

6.5.2 TensorFlow basics

6.6 Scikit learn

6.6.1 Features

6.6.2 Installation

6.6.3 Scikit modeling

6.6.4 Data representation in scikit

6.7 Keras.

6.7.1 Features

6.7.2 Building a model in Keras

6.7.3 Applications of Keras

6.8 Open NN

6.8.1 Application

6.8.2 RNN

6.9 Theano

6.9.1 An overview

6.10 Why go for Theano Python library?

6.10.1 PROS

6.10.2 CONS

6.11 Basics of Theano

6.11.1 Subtracting two scalars

6.11.2 Adding two scalars

6.11.3 Adding two matrices

6.11.4 Logistic function

7 Quality framework for explainable artificial intelligence (XAI) and machine learning applications

7.1 Introduction

7.2 Background

7.3 Integrated framework for AI applications development

7.4 AI systems characteristics vs. SE best practices

7.4.1 Explainable AI characteristics

7.5 ML lifecycle (model, data-oriented, and data analytics-oriented lifecycle)

7.6 AI/ML requirements engineering

7.7 Software effort estimation for AMD, RL, and NLP systems

7.7.1 Modified COCOMO model for AI, ML, and NLP applications and apps

7.8 Software engineering framework for AI and ML (SEF4 AI and ML) applications

7.9 Reference Architecture for AI &amp

ML

7.10 Evaluation of Reference Architecture (REF) for AI &amp

ML: explainable Chatbot case study

7.11 Conclusions and further research

8 Methods for explainable artificial intelligence

8.1 Preliminarily study

8.2 Importance of XAI for human-interpretable models

8.3 Overview of XAI techniques

8.4 Taxonomy of popular XAI methods

8.4.1 Backpropagation-based methods

8.4.2 Perturbation methods

8.4.3 Influence methods

8.4.4 Knowledge extraction

8.4.5 Concept methods

8.4.6 Visualization methods

8.4.7 Example-based explanation

8.5 Conclusion

9 Knowledge representation and reasoning (KRR)

9.1 Introduction

9.2 Methodology

9.2.1 Reference model

9.2.2 Ontologies

9.2.3 Knowledge graphs.

9.2.4 Semantic web technologies

9.2.5 ML

9.2.6 Tools and techniques

9.3 Results and discussion

9.3.1 Case study: using different techniques for representing medical knowledge [7]

9.3.2 Case study: using different techniques for representing academic knowledge [8]

9.3.3 Case study: using different techniques for representing farmer knowledge [9]

9.3.4 Case study: social media knowledge representation techniques [10]

9.3.5 Case study: using different techniques for representing cyber security knowledge [11]

9.4 Conclusion and future work

10 Knowledge visualization: AI integration with 360-degree dashboards

10.1 Introduction

10.2 Information visualization vs. knowledge visualization

10.3 Knowledge visualization in design thinking

10.4 Visualization in transferring knowledge

10.5 The knowledge visualization model

10.5.1 Knowledge visualization framework

10.6 Formats and examples of knowledge visualization

10.6.1 Conceptual diagrams

10.6.2 Visual metaphors

10.6.3 Knowledge animation

10.6.4 Knowledge maps

10.6.5 Knowledge domain visualization

10.7 Types and usage of knowledge visualization tools

10.8 Knowledge visualization templates

10.8.1 Mind maps

10.8.2 Swimlane diagrams

10.8.3 Matrix diagrams

10.8.4 Flowcharts

10.8.5 Concept maps

10.8.6 Funnel charts or diagrams

10.9 Visualization in machine learning

10.9.1 Decision trees

10.9.2 Decision graph

10.10 Conclusion

11 Empowering machine learning with knowledge graphs for the semantic era

11.1 Introduction

11.2 Tending towards digitally transformed enterprises

11.3 The emergence of KGs

11.4 Briefing the concept of KGs

11.5 Formalizing KGs

11.6 Creating custom KGs

11.7 Characterizing KGs

11.8 Use cases of KGs

11.9 ML and KGs.

11.10 KGs for explainable and responsible AI

11.11 Stardog enterprise KG platform

11.12 What CANNOT be considered a KG?

11.13 Conclusion

12 Enterprise knowledge graphs using ensemble learning and data management

12.1 Introduction

12.2 Current ensemble model learning

12.2.1 Bagging

12.2.2 Boosting

12.2.3 Random Forest

12.3 Related work and literature review

12.4 Methodology

12.4.1 Enhanced ensemble model framework

12.4.2 Training and testing datasets

12.4.3 Enhanced ensemble model and algorithm

12.5 Experimental setup and enterprise dataset

12.5.1 Ensemble models performance evaluation using enterprise knowledge graph

12.5.2 Tree classification as knowledge graph

12.6 Result and discussion

12.7 Conclusion

13 Illustrating graph neural networks (GNNs) and the distinct applications

13.1 Introduction

13.2 Briefing the distinctions of graphs

13.3 The challenges

13.4 ML algorithms

13.5 DL algorithms

13.6 The emergence of GNNs

13.7 Demystifying DNNs on graph data

13.8 GNNs: the applications

13.9 The challenges for GNNs

13.10 Conclusion

14 AI applications-computer vision and natural language processing

14.1 Object recognition

14.2 AI-powered video analytics

14.3 Contactless payments

14.4 Foot tracking

14.5 Animal detection

14.6 Airport facial recognition

14.7 Autonomous driving

14.8 Video surveillance

14.9 Healthcare medical detection

14.10 Computer vision in agriculture

14.10.1 Drone-based crop monitoring

14.10.2 Yield analysis

14.10.3 Smart systems for crop grading and sorting

14.10.4 Automated pesticide spraying

14.10.5 Phenotyping

14.10.6 Forest information

14.11 Computer vision in transportation

14.11.1 Safety and driver assistance

14.11.2 Traffic control.

14.11.3 Driving autonomous vehicles.

Notes:

Description based on publisher supplied metadata and other sources.

Part of the metadata in this record was created by AI, based on the text of the resource.

ISBN:

1-83724-425-1

1-5231-6305-4

1-83953-696-9

OCLC:

1410592840