In the rest of this book, we shall rely rather heavily on the vector space representations of data, because these arise naturally from measurement processes implemented in hardware and usable by robots. They also arise naturally as descriptions of animal nervous systems at the sensory level, where the incident energy or its logarithm tends to get coded in terms of neuronal spike frequencies.
We shall expand further on the details of the methods, metric, neural net and statistical, which have been outlined in the present chapter. We shall be concerned with issues in syntactic pattern recognition arising out of these methods and logically anterior to them, and we shall treat of the dynamical case of trajectories in the representation spaces and their recognition problems. But we shall not, for the reasons given, treat of the AI or fuzzy types of representation.