Automated Synthesis of Generalized Invariant Strategies via Counterexample-Guided Strategy Refinement

Authors: Kailun Luo, Yongmei Liu5800-5808

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We implemented our method and did experiments with a number of game problems. Our system can successfully synthesize solutions for most of the domains within a reasonable amount of time.
Researcher Affiliation Academia Kailun Luo,1,2 Yongmei Liu2 1 School of Cyberspace Security, Dongguan University of Technology, Dongguan 523808, China 2 Dept. of Computer Science, Sun Yat-sen University, Guangzhou 510006, China
Pseudocode Yes Algorithm 1: synthesize; Algorithm 2: refine; Algorithm 3: find Safe Region
Open Source Code No The paper states 'We implemented a prototype system using Python.' but does not provide any statement or link about the availability of the source code.
Open Datasets No The paper describes how game instances are generated from an 'initial database' and by using an SMT solver with a bound, but it does not use or provide access to a publicly available or open dataset in the traditional sense.
Dataset Splits No The paper does not specify any training, validation, or test dataset splits.
Hardware Specification Yes All experiments were conducted on a Mac OS machine with 2.30GHz CPU and 8GB RAM.
Software Dependencies No We use two SMT solvers for the first-order reasoning: Z3 [de Moura and Bjørner 2008] and CVC4 [Barrett et al. 2011]... We use the multi-agent model checker MCMAS [Lomuscio, Qu, and Raimondi 2017]... The paper lists software names but does not specify version numbers for Z3, CVC4, or MCMAS.
Experiment Setup Yes Considering the computational resources, we set the number c in Alg. 3 to 1, and the number k in Alg. 2 to 3. The time-out bound for the main algorithm is 3600 seconds. The time-out bound for the SMT solvers is 10 seconds.