Hybrid Evolutionary Programming Techniques
Bruno Buchberger, Witold Jacak.
In this research project, we want to develop “hybrid evolutionary programming techniques” for selected application areas of real world computing.
By “hybrid evolutionary programming” we mean the combination of symbolic computation techniques with recent evolutionary programming techniques, for example neural network and genetic programming, which share the common feature of self-adaption (evolution) in environments. In contrast to the pureluy evolutionary programming approach as pursued by most researchers in this area, in our hybrid approach, we integrate the benefits of advances symbolic computation techniques into the paradigm of evolutionary programming.
This integration is important because symbolic computation has proven quite successful for an impressive range of high-tech problem areas and it would be a mistake to discard these achievements when applying the new methods developed within the evolutionary programming paradigm. Often, in history, sticking to one approach only has been an obstacle to developing the full potential of new methods. In our view, the new evolutionary programming techniques can and should be combined with provenly successful symbolic methods in a number of different ways.