Hubness-Aware Classification, Instance Selection and Feature Construction: Survey and Extensions to Time-Series - Feature Selection for Data and Pattern Recognition - page 243

Table 11.4 Accuracy ± standard deviation (in %) for hubness-aware classifiers

Data set k NN AKNN hw- k NN NHBNN hFNN HIKNN

50words 72.4 ± 3.2 68.8 ± 3.3 • 71.3 ± 3.3 74.1 ± 2.7 76.4 ± 3.0 ⓦ 80.1 ± 2.9 ⓦ

Adiac 61.0 ± 4.1 58.6 ± 3.3 60.7 ± 4.0 63.8 ± 3.9 63.1 ± 3.8 65.9 ± 4.0 ⓦ

Beef 51.5 ± 15.6 31.8 ± 15.0 • 43.9 ± 14.3 50.4 ± 15.2 47.6 ± 15.6 47.8 ± 15.8

Car 71.7 ± 9.8 73.9 ± 9.4 76.0 ± 9.4 76.0 ± 8.6 76.8 ± 9.2 79.1 ± 8.7 ⓦ

CBF 100.0 ± 0.0 100.0 ± 0.0 100.0 ± 0.0 100.0 ± 0.0 100.0 ± 0.0 100.0 ± 0.0

Chlorine a

87.6 ± 1.3 68.7 ± 1.6 • 74.5 ± 1.6 • 68.7 ± 1.6 • 80.2 ± 1.4 • 91.8 ± 1.0 ⓦ

99.8 ± 0.2 99.6 ± 0.3 99.7 ± 0.2 99.6 ± 0.3 99.9 ± 0.1 99.9 ± 0.1

CinC-ECG-torso

Coffee 74.4 ± 15.9 68.7 ± 15.4 68.6 ± 15.1 71.2 ± 15.3 70.0 ± 15.2 76.9 ± 13.4

Cricket-X 78.0 ± 2.9 71.0 ± 2.9 • 77.9 ± 2.6 78.2 ± 3.0 79.3 ± 2.8 81.9 ± 2.6 ⓦ

± ± • ± ± ±

81.6 ± 2.9 ⓦ Cricket-Y 77.7 3.1 67.1 3.1 76.7 3.3 77.5 3.2 79.9 3.1

Cricket-Z 78.0 ± 3.1 70.9 ± 3.3 • 78.1 ± 3.1 78.6 ± 3.0 79.8 ± 3.0 82.2 ± 2.8 ⓦ

Diatom b 98.5 ± 1.6 100.0 ± 0.0 98.7 ± 1.4 99.1 ± 1.1 99.6 ± 0.8 99.6 ± 0.7

ECG200 85.9 ± 5.8 83.0 ± 4.9 85.3 ± 5.1 83.9 ± 4.6 85.2 ± 5.0 87.1 ± 5.0

ECGFiveDays 98.5 ± 0.7 97.7 ± 1.0 • 98.3 ± 1.0 98.3 ± 0.9 98.8 ± 0.7 99.0 ± 0.6

FaceFour 91.1 ± 6.3 92.2 ± 5.7 91.8 ± 6.7 94.0 ± 5.1 93.6 ± 5.4 94.1 ± 5.3

FacesUCR 96.4 ± 0.7 95.9 ± 0.8 96.7 ± 0.8 97.1 ± 0.7 97.5 ± 0.7 98.0 ± 0.6

FISH 77.3 ± 5.1 72.5 ± 5.7 • 77.1 ± 5.3 77.1 ± 5.5 77.9 ± 5.1 81.7 ± 4.5 ⓦ

Gun-Point 99.1 ± 1.4 95.4 ± 3.4 • 98.2 ± 2.0 97.9 ± 2.3 98.8 ± 1.6 99.0 ± 1.4

Haptics 50.5 ± 4.8 52.6 ± 5.5 51.3 ± 4.8 50.1 ± 4.8 51.9 ± 4.7 54.3 ± 4.6 ⓦ

InlineSkate 55.0 ± 3.8 44.8 ± 3.9 • 53.1 ± 3.7 53.6 ± 3.8 54.3 ± 4.0 60.1 ± 4.4 ⓦ

Italy c 96.5 ± 1.1 96.9 ± 1.1 96.7 ± 1.1 96.9 ± 1.0 96.9 ± 1.0 96.8 ± 1.0