Lyapunov Analysis of Neural Network Stability in an Adaptive Flight Control System

Yerramalla, Sampath; Fuller, Edgar; Mladenovski, Martin; Čukić, Bojan

doi:10.1007/3-540-45032-7_6

Cited by 26 publications

(22 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Important work that seems neglected by the scientific community is the stabilization science, which provides a common approach to studying system stability through stability analysis [1,6,20]. In this approach, a system is linearized around its operating point to determine a small-signal linearized model of that operating point.…”

Section: Improving Our Current Verification Toolsetmentioning

confidence: 99%

Safe Artificial Intelligence and Formal Methods

Vassev

2016

Leveraging Applications of Formal Methods, Verification and Validation: Foundational Techniques

View full text Add to dashboard Cite

In one aspect of our life or another, today we all live with AI. For example, the mechanisms behind the search engines operating on the Internet do not just retrieve information, but also constantly learn how to respond more rapidly and usefully to our requests. Although framed by its human inventors, this AI is getting stronger and more powerful every day to go beyond the original human intentions in the future. One of the major questions emerging along with the propagation of AI in both technology and life is about safety in AI. This paper presents the author's view about how formal methods can assist us in building safer and reliable AI.

show abstract

Section: Improving Our Current Verification Toolsetmentioning

confidence: 99%

Safe Artificial Intelligence and Formal Methods

Vassev

2016

Leveraging Applications of Formal Methods, Verification and Validation: Foundational Techniques

View full text Add to dashboard Cite

show abstract

“…al. developed a monitoring technique for the DCS neural network embedded in the IFCS (Yerramalla, Cukic, & Fuller, 2003a;Yerramalla, Fuller, & Cukic, 2003b) based on Lyapunov stability theory. The online monitors operate in parallel to the neural network with the goal of determining whether (or not), under given conditions, the neural network is convergent, meaning that all state transition trajectories converge to a stationary state.…”

Section: Rleted Work: Validating Online Adaptive Systemsmentioning

confidence: 99%

Validating neural network-based online adaptive systems: a case study

2007

Self Cite

View full text Add to dashboard Cite

Biologically inspired soft computing paradigms such as neural networks are popular learning models adopted in online adaptive systems for their ability to cope with the demands of a changing environment. However, continual changes induce uncertainty that limits the applicability of conventional validation techniques to assure the reliable performance of such systems. In this paper, we discuss a dynamic approach to validate the adaptive system component. Our approach consists of two run-time techniques: (1) a statistical learning tool that detects unforeseen data; and (2) a reliability measure of the neural network output after it accommodates the environmental changes. A case study on NASA F-15 flight control system demonstrates that our techniques effectively detect unusual events and provide validation inferences in a real-time manner.

show abstract

“…Discussing details relative to DCS algorithm and its properties is beyond the scope of this paper. For the description of online learning and DCS neural networks the reader is referred to [1][2][3][4][7][8][9]. Based on the existing literature describing DCS, the following basic definitions can be formulated.…”

Section: Dcsmentioning

confidence: 99%

Stability Monitoring and Analysis of Learning in Adaptive Systems

Yerramalla

Čukić

Campa

et al. 2006

2006 14th Mediterranean Conference on Control and Automation

Self Cite

View full text Add to dashboard Cite

The ability to assure reliability of adaptation is important in safety-critical applications. Traditional software V&V techniques cannot account for the time-evolving nature of a generic dynamic system, making them inapplicable for computing systems assurance. In this paper, we propose considering stability of adaptation as a heuristic measure of reliability, and present a stability monitoring system for detecting divergent learning behavior during online operation of adaptive systems. The monitoring system comprises of several Lyapunov-like functions that detect distinct states in learning that bifurcate away from stable behavior. Murphy's rule based on Dempster-Shafer theory is applied for combining stability information provided by individual monitors into an easily interpretable belief representation.The proposed analysis technique is evaluated using online learning experiments based on data generated by an actual adaptive flight control system. Results indicate that the stability monitoring system detects divercency conditions, and provides an insight into understanding whether the on-line learning converges to a stable state.

show abstract

Lyapunov Analysis of Neural Network Stability in an Adaptive Flight Control System

Cited by 26 publications

References 10 publications

Safe Artificial Intelligence and Formal Methods

Safe Artificial Intelligence and Formal Methods

Validating neural network-based online adaptive systems: a case study

Stability Monitoring and Analysis of Learning in Adaptive Systems

Contact Info

Product

Resources

About