zbMATH — the first resource for mathematics

Stochastic temporal logic abstractions: challenges and opportunities. (English) Zbl 06989596
Jansen, David N. (ed.) et al., Formal modeling and analysis of timed systems. 16th international conference, FORMATS 2018, Beijing, China, September 4–6, 2018. Proceedings. Cham: Springer (ISBN 978-3-030-00150-6/pbk; 978-3-030-00151-3/ebook). Lecture Notes in Computer Science 11022, 3-16 (2018).
Summary: Reasoning about uncertainty is one of the fundamental challenges in the real-world deployment of many cyber-physical system applications. Several models for capturing environment uncertainty have been suggested in the past, and these typically are parametric models with either Markovian assumptions on the time-evolution of the system, or Gaussian assumptions on uncertainty. In this paper, we propose a framework for creating data-driven abstractions of the environment based on stochastic temporal logics. Such logics allow combining the power of temporal logic-based absractions with powerful stochastic modeling techniques. Our framework allows constructing stochastic models using generalized master equations, which can be viewed as a nonparametric model capturing the dynamic evolution of the probabilities of system variables with time. Furthermore, we show how we can automatically infer temporal logic based abstractions from such a model. We give examples of applications for such a framework, and highlight some of the open problems and challenges in this approach.
For the entire collection see [Zbl 1396.68015].
68Qxx Theory of computing
Full Text: DOI