Online Incremental Learning for High Bandwidth Network Traffic Classification

Algorithm 2

Proposed algorithm.

(1) : flow instance

(2) : Nearest clusters from

(3) while new   do

(4)  merge to

(5) end while

(6) —Labeled Instance Injection (upon receiving)—

(7) if receive ()  then

(8)  - Inject

(9) end if

(10) —Micro-model Reconstruction (after one chunk)—

(11) : Number of clusters

(12) : Desired number of clusters

(13) while    do

(14)  delete clusters with timestamp

(15)  reduce clusters timestamp by

(16) end while

(17) gather all clusters in classification model

(18) reconstruct micro-model