Incremental Evolution in ANNs: Neural Netswhich Grow

摘要: This paper explains the optimisation of neural network topology using Incremental Evolutions that is, by allowing to expand adding its structure. method allows a grow from simple complex structure until it is capable fulfilling intended function. The approach somewhat analogous growth an embryo or evolution fossil line through time, therefore sometimes referred as embryology embryological algorithm. begins with general introduction, comparing this other competing techniques such Genetic Algorithm, Evolutionary Algorithms and Simulated Annealing. A literature survey previous work included, followed extensive new framework for application technique. Finally, examples applications discussion are presented.