Explainable Artificial Intelligence Based on Neuro-Fuzzy Modeling with Applications in Finance (Studies in Computational Intelligence, 964)

✍ Scribed by Tom Rutkowski

• Publisher: Springer
• Year: 2021
• Tongue: English
• Leaves: 175
• Category: Library

✦ Synopsis

The book proposes techniques, with an emphasis on the financial sector, which will make recommendation systems both accurate and explainable. The vast majority of AI models work like black box models. However, in many applications, e.g., medical diagnosis or venture capital investment recommendations, it is essential to explain the rationale behind AI systems decisions or recommendations. Therefore, the development of artificial intelligence cannot ignore the need for interpretable, transparent, and explainable models. First, the main idea of the explainable recommenders is outlined within the background of neuro-fuzzy systems. In turn, various novel recommenders are proposed, each characterized by achieving high accuracy with a reasonable number of interpretable fuzzy rules. The main part of the book is devoted to a very challenging problem of stock market recommendations. An original concept of the explainable recommender, based on patterns from previous transactions, is developed; it recommends stocks that fit the strategy of investors, and its recommendations are explainable for investment advisers.

✦ Table of Contents

Preface
Acknowledgments
Contents
List of Figures
List of Tables
1 Introduction
1.1 The Purpose of This Book
1.2 The Pursuit of Explainable Artificial Intelligence
1.3 Recommender Systems
1.4 Interpretability of Machine Learning Models
1.5 The Content and Main Results of the Book
References
2 Neuro-Fuzzy Approach and Its Application in Recommender Systems
2.1 Neuro-Fuzzy Systems as Recommenders
2.2 Fuzzy IF-THEN Rules and Learning Ability of the Recommenders
2.3 Interpretability and Explainability of Neuro-Fuzzy Recommenders
2.4 Rule Generation From Data
2.4.1 Input and Output Data for Neuro-Fuzzy Recommenders
2.4.2 Wang-Mendel Method of Rule Generation
2.4.3 Nozaki-Ishibuchi-Tanaka Method
2.5 Fuzzy IF-THEN Rules in Recommendation Problems
2.6 Classification in Recommenders
2.6.1 Neuro-Fuzzy Classifiers
2.6.2 One-Class Classifiers
2.6.3 Classification in Content-Based Recommender Systems
References
3 Novel Explainable Recommenders Based on Neuro-Fuzzy Systems
3.1 Recommender A
3.1.1 Feature Encoding
3.1.2 Description of the Proposed Recommender A
3.1.3 Systems Performance Evaluation
3.2 Recommender B
3.2.1 Introduction to the Proposed Recommender B
3.2.2 Description of the Recommender B
3.2.3 Criteria of Balance Evaluation Between Recommender Accuracy and Interpretability
3.2.4 Recommenders Performance Evaluation
3.2.5 Interpretability and Explainability of the Recommender
3.3 Recommender C
3.3.1 Nominal Attribute Values Encoding
3.3.2 Various Neuro-Fuzzy Systems as the Proposed Recommender C
3.3.3 Illustration of the Recommender Performance
3.4 Conclusions Concerning Recommenders A, B, and C
References
4 Explainable Recommender for Investment Advisers
4.1 Introduction to the Real-Life Application of the Proposed Recommender
4.2 Statement of the Problem
4.3 Description of the Datasets and Feature Selection
4.3.1 Data Enrichment and Dataset Preparation
4.3.2 Description of Selected Attributes—Simplified Version
4.3.3 Multidimensional Data Visualization
4.4 Definition of Fuzzy Sets
4.5 Fuzzy Rule Generation
4.6 Results of the System Performance
4.6.1 Recommendations Produced by the Recommender
4.6.2 Visualization of the Recommender Results
4.6.3 Explanations of the Recommendations
4.6.4 Evaluation of the Recommender Performance
4.7 Conclusions Concerning the Proposed One-Class Recommender
References
5 Summary and Final Remarks
5.1 Summary of the Contributions and Novelties
5.2 Future Research
5.3 Author's Contribution
References
Appendix A Description of Attributes—Full Version
Appendix B Fuzzy IF-THEN Rules
Appendix C Fuzzy Rules - Full Version
Appendix D Histograms of Attribute Values
Appendix E Fuzzy Sets for Particular Attributes
Appendix F Fuzzy Sets for Single Data Points