Guidance for the verification and validation of neural networks

Library	Item Barcode	Call Number	Material Type	Item Category 1	Status
Searching... PSZ JB	30000010214471	QA76.87 P85 2007	Open Access Book	Book	Searching... Unknown
Searching... PSZ KL	30000010214470	QA76.87 P85 2007	Open Access Book	Book	Searching... Unknown
Searching... PSZ KL	30000010214472	QA76.87 P85 2007	Open Access Book	Book	Searching... Unknown

This book provides guidance on the verification and validation of neural networks/adaptive systems. Considering every process, activity, and task in the lifecycle, it supplies methods and techniques that will help the developer or V&V practitioner be confident that they are supplying an adaptive/neural network system that will perform as intended. Additionally, it is structured to be used as a cross-reference to the IEEE 1012 standard.

Author Notes

Laura L. Pullum holds a B.S. in Mathematics, an M.S. in Operations Research, an M.B.A., and a D.Sc. in Systems Engineering and Operations Research.

Dr. Pullum has performed research and development in the dependable software areas of software fault tolerance, safety, reliability, and security for over 15 years. Dr. Pullum has written over 100 papers and reports on dependable software and has a patent (as co-inventor) in the area of fault tolerant agents. Dr. Pullum is a member of the IEEE Computer Society Technical Committee on Fault Tolerance Computing, IEEE Reliability and Software societies, Mensa, Women in Technology, and the U.S. Software System Safety Working Group.

050

Preface	p. vii
Acknowledgements	p. ix
1 Overview	p. 1
1.1 Definitions and Conventions	p. 2
1.2 Organization of the Book	p. 2
2 Areas of Consideration for Adaptive Systems	p. 5
2.1 Safety-Critical Adaptive System Example and Experience	p. 6
2.2 Hazard Analysis	p. 7
2.2.1 Development of a Neural Network Fault Model	p. 8
2.3 Requirements for Adaptive Systems	p. 12
2.4 Rule Extraction	p. 13
2.4.1 What is Rule Extraction?	p. 13
2.4.2 Rule Formats and Definitions	p. 14
2.4.3 Types of Rule Extraction	p. 14
2.4.4 How is Rule Extraction Useful in V&V?	p. 15
2.4.5 Advantages and Disadvantages	p. 15
2.5 Modified Life Cycle for Developing Neural Networks	p. 16
2.5.1 Nested Loop Model of Neural Network Development Process	p. 16
2.5.2 Safety Life Cycle for Hybrid Neural Networks	p. 18
2.6 Operational Monitors	p. 19
2.7 Testing Considerations	p. 20
2.7.1 Interface Testing	p. 21
2.7.2 Function Testing	p. 22
2.7.3 Knowledge Testing	p. 22
2.7.4 Structure Testing	p. 22
2.7.5 Neural Network Testing Tools	p. 23
2.8 Training Set Analysis	p. 24
2.8.1 Training Data with Too Many or Too Few Inputs	p. 24
2.8.2 "The Curse of Dimensionality"	p. 24
2.8.3 Data Redundancy	p. 24
2.8.4 Irrelevant Data	p. 24
2.8.5 Combining Different Data Sets into One	p. 25
2.8.6 Processing the Training Data	p. 25
2.8.7 Data Outliers	p. 25
2.8.8 Use of Rule Extraction/Insertion/Refinement with Training Data	p. 25
2.8.9 Training Data and Operational Monitoring	p. 25
2.8.10 Version Control of the Training Process	p. 26
2.9 Stability Analysis	p. 26
2.10 Configuration Management of Neural Network Training and Design	p. 27
2.11 Simulations of Adaptive Systems	p. 28
2.12 Neural Network Visualization	p. 29
2.13 Adaptive System and Neural Network Selection	p. 30
2.13.1 General Adaptive Systems	p. 30
2.13.2 Neural Network Systems at a High Level	p. 31
2.13.3 Neural Network Systems at a Low Level	p. 34
2.13.4 Neural Network Taxonomy	p. 36
3 Verification and Validation of Neural Networks - Guidance	p. 39
3.1 Process: Management	p. 40
3.1.1 Activity: Management of V&V	p. 40
3.2 Process: Acquisition	p. 43
3.2.1 Activity: Acquisition Support V&V	p. 43
3.3 Process: Supply	p. 45
3.3.1 Activity: Planning V&V	p. 45
3.4 Process: Development	p. 46
3.4.1 Activity: Concept V&V	p. 46
3.4.2 Activity: Requirements V&V	p. 52
3.4.3 Activity; Design V&V	p. 63
3.4.4 Activity: Implementation V&V	p. 84
3.4.5 Activity: Test V&V	p. 93
3.4.6 Activity: Installation and Checkout V&V	p. 97
3.5 Process: Operation	p. 101
3.5.1 Activity: Operation V&V	p. 101
3.6 Process: Maintenance	p. 107
3.6.1 Activity: Maintenance V&V	p. 107
4 Recent Changes to IEEE Std 1012	p. 111
Appendix A References	p. 119
Appendix B Acronyms	p. 123
Appendix C Definitions	p. 125

Available:*

On Order

Summary

Summary

Author Notes

Table of Contents