Learning Doubly Labeled Automata using Queries and Counterexamples

Sagar Chaki, Whitepaper, 2005.

Abstract: This whitepaper describes an extenstion of the L* algorithm for learning doubly labeled automata. A doubly labeled automaton (DLA) is a finite automaton whose states are labeled with sets of atomic propositions and whose transitions are labeled with actions. Thus a DLA is an amalgam of a Kripke structure and a finite state machine. The presented learning algorithm can be used to automate assume guarantee reasoning for checking trace containment between DLAs.