Learning Properties in LTL ∩ ACTL from Positive Examples Only

Ehlers, Rüdiger; Gavran, Ivan; Neider, Daniel

doi:10.34727/2020/isbn.978-3-85448-042-6_17

Cited by 3 publications

(3 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For learning models in LTL, several approaches have been proposed, leveraging SAT solvers (Neider & Gavran, 2018), automata (Camacho & McIlraith, 2019), and Bayesian inference (Kim et al, 2019). In fact, there are approaches for many temporal logics such as Property Specification Language (PSL) (Roy et al, 2020), Computational Tree Logic (CTL) (Ehlers et al, 2020) , Metric Temporal Logic (MTL) (Raha et al, 2023), etc.…”

Section: Related Workmentioning

confidence: 99%

Scarlet: Scalable Anytime Algorithms for Learning Fragments of Linear Temporal Logic

Raha,

Roy,

Fijalkow

et al. 2024

JOSS

Self Cite

View full text Add to dashboard Cite

In the past decade, incorporating data-driven AI techniques in system design has become mainstream in almost all branches of science and technology. Typically, systems powered by AI tend to be rather complex, far beyond human understanding. Naturally, one cannot always develop trust in such complex, so-called black-box systems, restricting their widespread use in safety-critical domains.To generate trust in systems, a standard approach in Explainable AI is to build simple explanations using human-understandable models. A number of recent works (Neider & Gavran, 2018) (Camacho & McIlraith, 2019) (Roy et al., 2020 have identified models in temporal logic to be both formal and explainable. Among temporal logics-Linear Temporal Logic (LTL)-arguably the most widely used temporal logic, has received particular focus due to its resemblance to natural language. Moreover, LTL is a de-facto standard in several fields of computer science, including model-checking, program analysis, and motion planning for robotics.Scarlet is one of the most competitive tools for learning explainable models in LTL (Raha et al., 2022). To learn such models, it relies on positive (or desirable) and negative (or undesirable) executions of the system under consideration. Based on the executions, it learns a concise model in LTL that is consistent with the given executions.

show abstract

Section: Related Workmentioning

confidence: 99%

Scarlet: Scalable Anytime Algorithms for Learning Fragments of Linear Temporal Logic

Raha,

Roy,

Fijalkow

et al. 2024

JOSS

Self Cite

View full text Add to dashboard Cite

show abstract

“…A number of different approaches have been proposed, leveraging SAT solvers [22], automata [5], and Bayesian inference [19], and extended to more expressive logics such as Property Specification Language (PSL) [25] and Computational Tree Logic (CTL) [10].…”

Section: State Of the Artmentioning

confidence: 99%

Scalable Anytime Algorithms for Learning Fragments of Linear Temporal Logic

Raha¹,

Roy²,

Fijalkow³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Linear temporal logic (LTL) is a specification language for finite sequences (called traces) widely used in program verification, motion planning in robotics, process mining, and many other areas. We consider the problem of learning LTL formulas for classifying traces; despite a growing interest of the research community existing solutions suffer from two limitations: they do not scale beyond small formulas, and they may exhaust computational resources without returning any result. We introduce a new algorithm addressing both issues: our algorithm is able to construct formulas an order of magnitude larger than previous methods, and it is anytime, meaning that it in most cases successfully outputs a formula, albeit possibly not of minimal size. We evaluate the performances of our algorithm using an open source implementation against publicly available benchmarks.

show abstract

“…State of the art. A number of different approaches have been proposed, leveraging SAT solvers [19], automata [4], and Bayesian inference [16], and extended to more expressive logics such as Property Specification Language (PSL) [24] and Computational Tree Logic (CTL) [9].…”

Section: Introductionmentioning

confidence: 99%

Scalable Anytime Algorithms for Learning Fragments of Linear Temporal Logic

Raha

Roy

Fijalkow

et al. 2022

Tools and Algorithms for the Construction and Analysis of Systems

Self Cite

View full text Add to dashboard Cite

Linear temporal logic (LTL) is a specification language for finite sequences (called traces) widely used in program verification, motion planning in robotics, process mining, and many other areas. We consider the problem of learning formulas in fragments of LTL without the $$\mathbf {U}$$ U -operator for classifying traces; despite a growing interest of the research community, existing solutions suffer from two limitations: they do not scale beyond small formulas, and they may exhaust computational resources without returning any result. We introduce a new algorithm addressing both issues: our algorithm is able to construct formulas an order of magnitude larger than previous methods, and it is anytime, meaning that it in most cases successfully outputs a formula, albeit possibly not of minimal size. We evaluate the performances of our algorithm using an open source implementation against publicly available benchmarks.

show abstract

Learning Properties in LTL ∩ ACTL from Positive Examples Only

Cited by 3 publications

References 27 publications

Scarlet: Scalable Anytime Algorithms for Learning Fragments of Linear Temporal Logic

Scarlet: Scalable Anytime Algorithms for Learning Fragments of Linear Temporal Logic

Scalable Anytime Algorithms for Learning Fragments of Linear Temporal Logic

Scalable Anytime Algorithms for Learning Fragments of Linear Temporal Logic

Contact Info

Product

Resources

About