Generating Real-time Decision Systems with the new miAamics

We sketch miAamics, an approach and a tool to rapidly evaluate large systems of rules. These large systems of rules can be used to express performance critical decision functions and allow for the miAamics approach to optimize the function and to generate its implementation fully automatically. In this way, we allow experts to define functions without having to be familiar with general purpose programming languages and also allow to optimize existing decision functions that can be expressed in form of these rules. The proposed approach first transforms the system of rules to Algebraic Decision Diagrams. From this data structure, we generate code in a variety of commonly used target programming languages. We present preliminary results from experiments with randomly generated rules and show that the proposed representation is significantly faster to evaluate and is also smaller in size than the original representation. We give an outlook on possible applications for the miAamics approach to real world tasks focusing on the field of machine learning. In particular, we aim to reduce ensembles of classifiers and to allow for a much faster evaluation of these classification methods.