38 Knowledge Discovery from Data Streams
Dynamic environments with non-stationary distributions require the for-
getfulness of the observations not consistent with the actual behavior of the
nature. Drift detection algorithms must not only adapt the decision model
to newer information but also forget old information. The memory model
also indicates the forgetting mechanism. Weighting examples corresponds to a
gradual forgetting. The relevance of old information is less and less important.
Time windows correspond to abrupt forgetting (weight equal 0). The exam-
ples are deleted from memory. We can combine, of course, both forgetting
mechanisms by weighting the examples in a time window (see Klinkenberg
(2004)).
3.2.2.2 Detection Methods
The Detection