Error-awareness Accelerates Active Automata Learning
This work addresses scalability issues in AAL for systems with many error-prone inputs, offering significant speedups for model learning in domains like software verification, though it is incremental as it builds on existing algorithms.
The paper tackles the challenge of scaling active automata learning (AAL) algorithms to larger models, especially when many inputs cause errors, by adapting the L# algorithm to leverage observable errors and varying levels of domain knowledge about non-error inputs, resulting in learning acceleration by orders of magnitude with strong knowledge to a single order with limited knowledge.
Active automata learning (AAL) algorithms can learn a behavioral model of a system from interacting with it. The primary challenge remains scaling to larger models, in particular in the presence of many possible inputs to the system. Modern AAL algorithms fail to scale even if, in every state, most inputs lead to errors. In various challenging problems from the literature, these errors are observable, i.e., they emit a known error output. Motivated by these problems, we study learning these systems more efficiently. Further, we consider various degrees of knowledge about which inputs are non-error producing at which state. For each level of knowledge, we provide a matching adaptation of the state-of-the-art AAL algorithm L# to make the most of this domain knowledge. Our empirical evaluation demonstrates that the methods accelerate learning by orders of magnitude with strong but realistic domain knowledge to a single order of magnitude with limited domain knowledge.