Welcome to RERS - a Challenge in Active Learning!

With RERS we want to establish a community of researchers and practitioners interested in the practical application of automata learning technology.

RERS stands for Regular Extrapolation of Reactive Systems. We chose the term Regular Extrapolation to indicate that in practice we will not be able to fully infer the behaviour of the target system, which often will not be regular anyway. Rather we are able to construct in some sense optimal regular images/views of these behaviours. In fact, characterizing these kinds of views e.g. by abstraction techniques is one of the major challenges of RERS.

There are multiple dimensions to RERS, like the design of problems specific model structures, the development of flexible abstraction technologies, and the adequate treatment of parameters and values treatment. These need to be related to learning-specific technologies for e.g. treating counter examples, realizing equivalence queries, or exploiting domain-specific knowledge about the target systems, and, a potential show stopper in practice, to adequate testing technology guaranteeing supporting the required kind of querying.

We envisage to establish targeted sub-communities specialized and interested in specific scenarios. One such scenario could be e.g. the `classical´ learning scenario as addressed e.g. by the Zulu challenge, but there are many more, each of them with a particular learning profile. Examples range from the learning of I/O automata or Mealy machines for complex reactive systems or protocols, over the construction of(abstract) behavioural models of software components e.g.as a means to automate assume guarantee reasoning, to the detection of business processes by real life observation.

This short sketch is only a beginning, and we are inviting you to join, and contribute and extend this vision together! In case you are interested, please fill out the participation form.