Symbolic machine learning refers to techniques that do not rely on neural networks or other sub-symbolic approaches. Examples include version spaces, k-means and decisoon trees. It is unclear where techniques such as geberic algorithms and swarm computing belong, however a good rule of thumb is to look at the kinds of output rules they produce. If a genetic algorithm has a massive set of paaremeters that are being manipulated, then this is sub-symbolic, but of the outcome is. relatvely simple set of decison rules, then it feels more symbolic.
Used on Chap. 18: page 439