Abstract
In the paper algorithms are considered which allow to optimize decision trees consecutively againsts relatively different criterions. For decision tables over an arbitrary infinite restricted information system [4], these algorithms have polynomial time complexity.