Figure3. Offline KNN learning scheme
Figure 3 depicts a standard simplified offline algorithm scheme.
Online K-means Algorithm
Compared with offline methods of learning, online learning algorithms process an endless number (xi) and
outcome (yi) of potentially predictors before they meet each other. The YT is computed as y't = ht-1, as the
goal of learning is the prediction of the present input yt by using the model HT-1 that is already known (xt-1).
Another distinction is that the training and analysis data sets are not entirely isolated, but that a sample of
cases is used for model testing until it is used for model training. Figure 4 depicts a typical online forecasting
mark scheme as follows:
y’t = ℎ
t−1
(𝑥
t−1
).
(3)
Do'stlaringiz bilan baham: |