A conservative bound for the probability of failure of a 1-out-of-2 protection system with one hardware-only and one software-based protection train

Bishop, Peter; Bloomfield, Robin; Littlewood, Bev; Popov, Peter; Povyakalo, Andrey; Strigini, Lorenzo

doi:10.1016/j.ress.2014.04.002

Cited by 8 publications

(7 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the unknown ground truth over the underlying data distribution, this approach usually requires to either make an assumption over the distribution [22] or learn the distribution [26]. Note that, the obtained probabilistic assessment can be bounded [2,12]. One step further, it might be interesting to understand, and verify, the safety and reliability issues in learning-enabled autonomous systems where deep learning components interact with symbolic or probabilistic components [9,20].…”

Section: Probabilistic Assessment and Safety Argumentmentioning

confidence: 99%

Safety and reliability of deep learning

Huang

2021

Proceedings of the 1st International Workshop on Verification of Autonomous &Amp; Robotic Systems

View full text Add to dashboard Cite

Robotics and Autonomous Systems (RAS) become ever more relying on deep learning components to support their perception and decision making. Given RAS will inevitably be applied to safety critical applications, efforts are needed to ensure that the deep learning is safe and reliable. In this lecture, I will give a brief overview on recent progress in the verification and validation techniques for deep learning, focusing on two major safety and reliability risks, i.e., robustness and generalisation. We consider formal verification, statistical evaluation, reliability assessment, and runtime monitoring techniques, all of which complement with each other in providing assurance to the reliability of deep learning in operation. The challenges and future directions will also be discussed.

show abstract

Section: Probabilistic Assessment and Safety Argumentmentioning

confidence: 99%

Safety and reliability of deep learning

Huang

2021

Proceedings of the 1st International Workshop on Verification of Autonomous &Amp; Robotic Systems

View full text Add to dashboard Cite

show abstract

“…Appendix A.1 shows that there is a conservative, linear approximation to the confidence region, D 1 , that can be used to derive an analytic upper bound q u > q s based on the bin-filling algorithm reported in [20,21]. This analysis shows that the upper bound q u can always be obtained by assigning all the failure probability to the cutset with the smallest total of component tests.…”

Section: Structure-based Confidence Bound Modelmentioning

confidence: 99%

“…Since equation (A.5) is linear, finding the conservative confidence bound (A.6) is a linear programming problem, which can be solved with the bin-filling algorithm, reported in [20,21]:…”

Section: Appendix a Confidence Bound Approximationmentioning

confidence: 99%

A conservative confidence bound for the probability of failure on demand of a software-based system based on failure-free tests of its components

Bishop

Povyakalo

2020

Reliability Engineering & System Safety

Self Cite

View full text Add to dashboard Cite

show abstract

“…Recent work has estimated a conservative bound for the failure probability of a program-pair in a special case [36,37], but a general solutions are not available The EL and LM models attempt to describe what is true 'on average' for a large population of programs. Subsequently, an online collection of programs has been used for demonstrating the LM model [21].…”

Section: Safety Integrity Levelmentioning

confidence: 99%

A Diversity Model Based on Failure Distribution and its Application in Safety Cases

Chen

May

2016

IEEE Trans. Rel.

View full text Add to dashboard Cite

Abstract-This work develops a new basis for evaluating the reliability benefits of diverse software, based on fault injection testing. In particular, the work investigates new forms of argumentation that could in principle be used to justify diversity as a basis for the construction of safety claims. Failure distributions of two versions of diverse software under various fault conditions are revealed separately by fault injection methods, and then the common failure probability of the version-pair can be estimated. The approach is justified theoretically, and cross validated with other work. This method is also used to explain the fundamental influence of failure distributions on diversity. Furthermore, the unique capabilities of the method are demonstrated by implementation of the fault injection test on a program pair Index Terms-Software Diversity, Safety Case, Safety Critical System, Fault Injection, Multi-version, Reliability. Abbreviations: AFRAverage failure rate CFP Common failure probability EL Eckhardt and Lee FD Failure distribution FI Fault injection IFP s-Independent failures model LM Littlewood and Miller MR Meulen and Revilla MSF Mmodified sign function NVP New version-pair RI Reliability improvement SFP

show abstract

A conservative bound for the probability of failure of a 1-out-of-2 protection system with one hardware-only and one software-based protection train

Cited by 8 publications

References 9 publications

Safety and reliability of deep learning

Safety and reliability of deep learning

A conservative confidence bound for the probability of failure on demand of a software-based system based on failure-free tests of its components

A Diversity Model Based on Failure Distribution and its Application in Safety Cases

Contact Info

Product

Resources

About