Differential evolution versus genetic algorithms: towards symbolic aggregate approximation of non-normalized time series

The differential evolution (DE) is a very powerful search method for solving many optimization problems. In this paper we present a new scheme (DESAX) based on the differential evolution to localize the breakpoints utilized with the symbolic aggregate approximation method; one of the most important symbolic representation techniques for times series data. We compare the new scheme with a previous one (GASAX), which is based on the genetic algorithms, and we show how the new scheme outperforms the original one. We also show how (DESAX) can be used for the symbolic aggregate approximation of non-normalized time series.