mannprerak2 / outlier_detection

DWDM Project submission on A fast distance based outlier detection using DHCA.

A fast distance based outlier detection technique.

Demo Results -

Human Gene -

Running dhca outlier detector:
Outlier Labels: ['11_1719', '2_1460', '1_445']
Outlier Indexes: [28, 153, 93]
Outlier Scores [1249.6666666666667, 850.3333333333334, 834.3333333333334]
Verified: 6 out of 311
Calculations: 2896 out of 48205
Running Time: 3.18 seconds

Corel Dataset -

Outlier Labels: ['58685', '64221', '4271', '49781', '65648', '56685', '42144', '42156', '56756', '18262']
Outlier Indexes: [58684, 64220, 4270, 49780, 65647, 56684, 42143, 42155, 56755, 18261]
Outlier Scores [4.247031512521615, 3.6606322652017136, 3.3741147618420184, 3.2723253978677374, 3.2630320559384436, 3.181449975709505, 3.0844966599382495, 3.038401973204192, 2.9722768448265517, 2.938333571367344]
Verified: 275 out of 68040
Calculations: 20945704 out of 2314720800 i.e (N*N)/2
Running Time: 291.33 seconds