An experimental comparison of symbolic and neural learning algorithms

Comparative strengths and weaknesses of symbolic and neural learning algorithms are analysed. Experiments comparing the new generation symbolic algorithms and neural network algorithms have been performed using twelve large, real-world data sets. Results indicate that their performances are comparable for most of the different data sets. However, in some data sets neural network algorithms´ predicted accuracies are statistically significant than symbolic algorithms and in others symbolic algorithms´ performances are superior. In general, neural network algorithms are found quite robust when noisy and missing data are introduced in the data sets