Shifting Automata Learner

This is a project to implement an algorithm that can learn shifting automata. Shifting implies changes in information a Learner receives from a Teacher during the process of learning. This work is based primarily on Dana Angluin's paper "Learning Regular Sets From Queries and Counterexamples." It uses the algorithm described in Kearns' and Vazirani's textbook "An Introduction To Computational Learning Theory."

Acquiring

Download the latest stable version from sourceforge. Latest release version is 0.2. The project is in beta stage.

Source code can be browsed in the repository. Read instructions if you would like to check out the latest source for modification, building, and using code in your projects.

Using

In order to compile and run the application, complete the following steps:

1. Get source code either from a released archive or source code repository.
2. Run

   ant jar

   from the project directory that contains the file named build.xml. Ant is a build tool tailored to Java applications.
3. To start the application, execute

   java -jar dfa-learner.jar

   from the same directory. At this time, the application will:
   1. Prompt user for options to be passed to the algorithm
   2. Graphically display the hypothesis automaton and the classification tree

Descriptions of possible command line options are in the file OPTIONS. Usage instructions are in the file README.

If you find a problem in the application, or would like to suggest/implement a feature, please file a bug.

Authors

The project was created and is maintained by Rouslan Solomakhin under advisement of Christino Tamon. This page was last updated on January 4th, 2007.