Active Semisupervised Clustering Algorithm with Label Propagation for Imbalanced and Multidensity Datasets

Algorithm 1

Selecting data by using MST clustering algorithm 𝑆𝑒𝑙𝑒𝑐𝑡𝐷𝑎𝑡𝑎𝑃𝑜𝑖𝑛𝑡𝐷,𝑝.

(1)   Let 𝑚=|𝐷|×𝑝, 𝑚 is the number of data points

    to be selected, |𝐷| is the size of dataset 𝐷.

(2)   Use Prime method to construct MST of 𝐷.

(3)   Foreach  edge in MST  do

(4)  Compute edge’s inconsistent value 𝑓.

(5)   End Foreach

(6)   Sort all edges in descending order according to 𝑓.

(7)   Insert the sorted edges into a list: edgesLst.

(8)  Foreach  edge in edgesLst do

(9)    Delete edge from MST

(10)  Check the number of partitions in MST, num

(11)  If  num == m then

(12)   Generate num clusters 𝑇₁,𝑇₂,…,𝑇_𝑚 from MST

(13)    Break

(14) End If

(15) End Foreach

(16) Foreach cluster T in 𝑇₁,𝑇₂,…,𝑇_𝑚  do

(17)  Compute density of each point in T

(18)  Select one data with max density and add it to 𝐷_𝑙

(19) End Foreach

(20)  Query oracle about labels of data in 𝐷_𝑙.

(21) Return 𝑇₁,𝑇₂,…,𝑇_𝑚 and 𝐷_𝑙.