TAG: Learning Timed Automata from Logs

Event logs are often one of the main sources information to understand behavior a system. While numerous approaches have extracted partial from event logs, in this work, we aim at inferring global model system its logs. We consider real-time systems, which can be modeled with Timed Automata: our approach is thus Automata learner. There handful related however, they might require lot parameters or produce that either undeterministic lack precision. In contrast, proposed approach, called TAG, requires only parameter and learns deterministic Automaton having good tradeoff between accuracy complexity automata. This allows getting an interpretable accurate considered. Our experiments compare work demonstrate merits.