 |
To describe our prediction model, we assume that the data point, we assume data domains.
And t and i are the relative time in compressed periods (ultimately decomposed block in our case.
It also requires a parameter, еш (i) , which characterizes data points in the compressed period.
Each compressed period i has a series of data points during time t.
We next define the data profile (еш (i) ), which depends on whether the block under consideration is associated with data of typical or abnormal behavior.
еш (i) is characterizing data points for each block i.
|
IEEE/ICACT20220021 Slide.06
[Big Slide]