Decision Trees, Tables, and Lattices

The use of small built-in processors in pattern recognition machines is becoming much more widespread, with the advent of very cheap processing power in the form of LSI microprocessors and other devices. It is therefore quite reasonable to design systems which evaluate rather complicated features according to the dictates of an “intelligent” computer program and arrive at a decision by branching and sequential feature extraction, rather than by deriving several features in parallel and then classifying by hypersurface separation.