Example global attributes

For a given channel, we could define the global maximum $\mathit{max}$ for a given channel as

$$\mathit{max}(c) = m\ \mathrm{s.t.}\ \forall t\ c[t] \leq m, t \in \mathrm{domain}(c) \wedge \exists t\ \mathrm{s.t.}\ c[t] = m$$

Similarly, we could define a distance measure on a single channel as:

$$\mathit{dist}(c) = \sqrt{ \sum_{t=1}^{t_{max}} c[t]^2 - c[t-1]^2 }$$

These global features can be useful for classification; in fact, in the past this has been one of the most popular means of ad-hoc feature extraction technique in temporal classification domains. However, they are usually not sufficient for doing high-accuracy classification. In addition, used alone, they do not usually lead to the production of useful human-readable descriptions of the learnt concept.