Uncertainty propagation using probabilistic affine forms and concentration of measure inequalities

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We consider the problem of reasoning about the probability of assertion violations in straight-line, nonlinear computations involving uncertain quantities modeled as random variables. Such computations are quite common in many areas such as cyber-physical systems and numerical computation. Our approach extends probabilistic affine forms, an interval-based calculus for precisely tracking how the distribution of a given program variable depends on uncertain inputs modeled as noise symbols. We extend probabilistic affine forms using the precise tracking of dependencies between noise symbols combined with the expectations and higher order moments of the noise symbols. Next, we show how to prove bounds on the probabilities that program variables take on specific values by using concentration of measure inequalities. Thus, we enable a new approach to this problem that explicitly avoids subdividing the domain of inputs, as is commonly done in the related work. We illustrate the approach in this paper on a variety of challenging benchmark examples, and thus study its applicability to uncertainty propagation.

Original languageEnglish
Title of host publicationTools and Algorithms for the Construction and Analysis of Systems - 22nd International Conference, TACAS 2016 and Held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2016, Proceedings
EditorsJean-François Raskin, Marsha Chechik
PublisherSpringer Verlag
Pages225-243
Number of pages19
ISBN (Print)9783662496732
DOIs
Publication statusPublished - 1 Jan 2016
Externally publishedYes
Event22nd International Conference on Tools and Algorithms for the Construction and Analysis of Systems, TACAS 2016 and held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2016 - Eindhoven, Netherlands
Duration: 2 Apr 20168 Apr 2016

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9636
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference22nd International Conference on Tools and Algorithms for the Construction and Analysis of Systems, TACAS 2016 and held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2016
Country/TerritoryNetherlands
CityEindhoven
Period2/04/168/04/16

Fingerprint

Dive into the research topics of 'Uncertainty propagation using probabilistic affine forms and concentration of measure inequalities'. Together they form a unique fingerprint.

Cite this