Showing 1–1 of 1 results for author: Manta, A

Semantic Labelling and Learning for Parity Game Solving in LTL Synthesis

Authors: Jan Křetínský, Alexander Manta, Tobias Meggendorfer

Abstract: We propose "semantic labelling" as a novel ingredient for solving games in the context of LTL synthesis. It exploits recent advances in the automata-based approach, yielding more information for each state of the generated parity game than the game graph can capture. We utilize this extra information to improve standard approaches as follows. (i) Compared to strategy improvement (SI) with random i… ▽ More We propose "semantic labelling" as a novel ingredient for solving games in the context of LTL synthesis. It exploits recent advances in the automata-based approach, yielding more information for each state of the generated parity game than the game graph can capture. We utilize this extra information to improve standard approaches as follows. (i) Compared to strategy improvement (SI) with random initial strategy, a more informed initialization often yields a winning strategy directly without any computation. (ii) This initialization makes SI also yield smaller solutions. (iii) While Q-learning on the game graph turns out not too efficient, Q-learning with the semantic information becomes competitive to SI. Since already the simplest heuristics achieve significant improvements the experimental results demonstrate the utility of semantic labelling. This extra information opens the door to more advanced learning approaches both for initialization and improvement of strategies. △ Less

Submitted 22 July, 2019; originally announced July 2019.