5 lines
192 B
Plaintext
5 lines
192 B
Plaintext
|
kNN algorithm can also be used for AD
|
||
|
|
if the k closest point is further away, a sample is considered more anomalous
|
||
|
|
$r=\frac{k}{2N\cdot pdf}$
|
||
|
|
Powerful method, as it can model the pdf directly
|