Untold volumes of material have been written on the topic of signal preprocessing for prediction and classification. The treatment here will not pretend to do more than lightly scratch the surface of this huge topic. In fact, we will not even present a survey of techniques. You are invited to peruse other references for specialized topics.
Rather, this chapter will present several specific methods for preprocessing time series in order to compute variables that make suitable inputs to deep belief nets (and, indeed, many other models). The methods shown here ...
