Application of Neural Networks in High Assurance Systems: A Survey

  title={Application of Neural Networks in High Assurance Systems: A Survey},
  author={Johann M. Ph. Schumann and Pramod Gupta and Yan Liu},
  booktitle={Applications of Neural Networks in High Assurance Systems},
Artificial Neural Networks (ANNs) are employed in many areas of industry such as pattern recognition, robotics, controls, medicine, and defence. Their learning and generalization capabilities make them highly desirable solutions for complex problems. However, they are commonly perceived as black boxes since their behavior is typically scattered around its elements with little meaning to an observer. The primary concern in safety critical systems development and assurance is the identification… 

Hardening of Artificial Neural Networks for Use in Safety-Critical Applications - A Mapping Study

It is envisioned that future software engineering will need to focus on further investigating these methods and increasing the maturity and understanding of existing approaches, with the goal to develop clear guidance for proper engineering of high-quality ANNs.

The assurance of Bayesian networks for mission critical systems

A framework for conceptualising and communicating the distinctions between BNSs and conventional software systems is proposed and an assurance-focussed BNS analysis technique is introduced that can provide targeted information on mission-critical aspects of a BNS.

Toward Certification of Machine-Learning Systems for Low Criticality Airborne Applications

This paper analyzes the current airborne certification standards and shows that all objectives of the standards can be achieved for a low-criticality ML-based system if certain assumptions about ML development workflow are applied.

Using Machine Learning Safely in Automotive Software: An Assessment and Adaption of Software Process Requirements in ISO 26262

A detailed assessment and adaption of ISO 26262 for ML is done, specifically in the context of supervised learning, to address a conflict between the need to innovate and theneed to improve safety in automotive development.

Hybrid Decompositional Verification for Discovering Failures in Adaptive Flight Control Systems

This paper approaches the problem of adaptive control methodologies with a variant of compositional verification, and finds that composition of component inputs throughout the system leads to overall system test vectors that may elucidate the undesirable behavior.

A survey on artificial intelligence assurance

This manuscript provides a systematic review of research works that are relevant to AI assurance, between years 1985 and 2021, and aims to provide a structured alternative to the landscape.

Explicit Image Quality Detection Rules for Functional Safety in Computer Vision

It is hypothesised that low-level and primitive image analysis driven by explicit rules facilitates complying with safety standards, which improves the real-world applicability of existing proposed solutions.

Safely Entering the Deep: A Review of Verification and Validation for Machine Learning and a Challenge Elicitation in the Automotive Industry

The state-of-the-art in verification and validation of safety-critical systems that rely on machine learning is reviewed, confirming that ISO 26262 largely contravenes the nature of DNNs.

Increasing Safety by Combining Multiple Declarative Rules in Robotic Perception Systems

This paper presents an evaluation of different methods for modelling combinations of simple explicit computer vision rules designed to increase the trustworthiness of the perception system, finding that it is possible to improve the safety of the system with some performance cost, depending on the acceptable risk level.



Validating a neural network-based online adaptive system

This research investigated the V&V of neural network-based online adaptive systems and developed a novel validation approach consisting of two important methods, and defined a confidence measure, the validity index, to validate the predictions of the Dynamic Cell Structure (DCS) network in IFCS.

Performance Estimation of a Neural Network-Based Controller

A dynamic approach to estimate the performance of two types of neural networks employed in an adaptive flight controller: the validity index for the outputs of a Dynamic Cell Structure (DCS) network and confidence levels for the output of a Sigma-Pi (or MLP) network.

Performance Monitoring and Assessment of Neuro-Adaptive Controllers for Aerospace Applications Using a Bayesian Approach

Modem aircraft, UAVs, and robotic spacecraft pose substantial requirements on controllers in the light of ever increasing demands for reusability, affordability, and reliability. The individual

Tools and Methods for the Verification and Validation of Adaptive Aircraft Control Systems

Several advanced methods proposed for verification and validation (V&V) of adaptive control systems, including Lyapunov analysis, statistical inference, and comparison to the well-known Kalman filters are presented.

Industrial use of safety-related artificial neural networks

Neural network products are actively being marketed and some are routinely used in safetyrelated areas, including cancer screening and fire detection in office blocks. Some are medical devices

Foundation for neural network verification and validation

  • G. Peterson
  • Computer Science
    Defense, Security, and Sensing
  • 1993
Some clarifying concepts related to evaluation, and a process for developing neural networks in which the role of evaluation is emphasized are proposed.

Adaptive control software: can we guarantee safety?

This work applied a nonconventional V&V approach to an adaptive flight control system that employs neural network learning for online adaptation and found it to be suitable for online adaptive systems.

Intelligent systems in the automotive industry: applications and trends

This paper provides an overview and a sampling of many of the ways that the automotive industry has utilized AI, soft computing and other intelligent system technologies in such diverse domains like manufacturing, diagnostics, on-board systems, warranty analysis and design.

A deployed engineering design retrieval system using neural networks

A neural information retrieval system (NIRS) is described, now in production within the Boeing Company, which has been developed for the identification and retrieval of engineering designs, potentially saving large amounts of nonrecurring costs.

Application of Neural Networks in Power Systems; A Review

According to the growth rate of NNs application in some power system subjects, a brief overview in fault diagnosis, security assessment, load forecasting, economic dispatch and harmonic analyzing is introduced.