We studied the learning algorithm for reversible languages.
Beginning with the prefix automaton for a data set, we merged offending states until the resulting machine was reversible.
We saw that a particular finite set of sentences (encoding states and transitions between them) sufficed for the algorithm to hypothesize the target machine. This was a tell-tale set in Angluin’s sense.
- homework
- Run Angluin’s reversible learner on the following data set:
- aaab
- aacc
- ac
- abac