Working demo of Lernd

No functionality changes from the last version.

In this version, almost everything from the original paper is implemented and there are 2 demo programs ("predecessor", and "even"). Things not implemented yet are:

• splitting (B, P, N) tuple into multiple worlds for better generalization in training;
• iterative deepening - automatically searching for the best hyper-parameters (rule templates).

Working demo of Lernd

In this version, almost everything from the original paper is implemented and there are 2 demo programs ("predecessor", and "even"). Things not implemented yet are:

• splitting (B, P, N) tuple into multiple worlds for better generalization in training;
• iterative deepening - automatically searching for the best hyper-parameters (rule templates).