Advances in Artificial Intelligence

Research Article

Twin Support Vector Machine for Multiple Instance Learning Based on Bag Dissimilarities

Table 3

Performance comparison of nonlinear MIL-TWSVM with different dissimilarity score.


Datasets	Max-Hausdorff , Acc ± std (%)	Min-Hausdorff , Acc ± std (%)	Euclidean (average) , Acc ± std (%)	City Block , Acc ± std (%)	Chi-squared , Acc ± std (%)	Mahalanobis , Acc ± std (%)	EMD , Acc ± std (%)

Musk 1	10⁻², 2³ 55.57 ± 0.0	10⁰, 2¹ 55.57 ± 0.0	10⁻², 2² 55.57 ± 0.0	10³, 2² 55.57 ± 0.0	10⁻², 2¹ 78.89 ± 3.68	10⁻⁴, 2³ 71.14 ± 5.22	10⁻¹, 2⁰ 55.57 ± 0.0
Musk 2	10⁻², 2³ 58.25 ± 3.45	10⁻³, 2⁵ 55.57 ± 0.0	10⁻³, 2³ 57.03 ± 2.4	10⁻³, 2² 55.57 ± 0.0	10⁰, 2¹ 77.78 ± 4.04	10⁻², 2⁴ 77.70 ± 3.26	10⁻², 2⁰ 55.57 ± 0.0
Mutagenesis-atoms	10⁻², 2¹ 85.52 ± 1.28	10⁻¹, 2⁴ 81.58 ± 5.88	10⁻³, 2³ 68.42 ± 0.0	10⁻², 2³ 68.42 ± 0.0	10⁻¹, 2⁵ 68.42 ± 0.0	10⁻², 2³ 68.42 ± 0.0	10⁻³, 2⁰ 89.47 ± 4.07
Winter wren	10⁻³, 2³ 97.56 ± 4.67	10⁻⁴, 2⁴ 98.02 ± 2.2	10⁻⁴, 2³ 91.10 ± 3.95	10⁻¹, 2⁵ 82.86 ± 5.69	10⁻⁵, 2⁶ 79.92 ± 3.05	10⁻³, 2³ 87.18 ± 4.84	10⁻³, 2⁴ 95.49 ± 4.66
Brown creeper	10⁻³, 2¹ 97.27 ± 3.84	10⁻⁵, 2³ 97.89 ± 3.71	10⁻³, 2⁰ 92.30 ± 4.08	10⁻⁵, 2² 85.8 ± 4.53	10⁻³, 2⁰ 81.10 ± 3.26	10⁻³, 2¹ 90.03 ± 3.35	10⁻⁴, 2⁵ 96.72 ± 3.24
Elephant	10⁻⁵, 2³ 83.62 ± 3.37	10⁻⁷, 2⁵ 83.33 ± 5.08	10⁻¹, 2⁰ 50 ± 0.0	10⁰, 2⁰ 50 ± 0.0	10⁰, 2³ 55 ± 0.0	10⁰, 2⁰ 50 ± 0.0	10⁻¹, 2⁰ 50 ± 0.0
Fox	10⁻⁷, 2⁴ 63.89 ± 1.01	10⁻², 2³ 75 ± 4.49	10⁻⁵, 2⁴ 61.11 ± 6.98	10⁻⁵, 2¹ 57.78 ± 7.85	10⁻⁵, 2⁰ 55 ± 0.0	10⁻⁴, 2¹ 54 ± 4.76	10⁻³, 2¹ 50 ± 0.0
Tiger	10⁻⁵, 2³ 85.63 ± 1.22	10⁻⁵, 2³ 83.13 ± 6.02	10¹, 2¹ 50 ± 0.0	10¹, 2¹ 50 ± 0.0	10⁻², 2² 50.5 ± 2.06	10⁰, 2¹ 77 ± 8.12	10⁰, 2¹ 50 ± 0.0
eastWest	10⁻³, 2⁴ 80 ± 24.49	10⁰, 2¹ 80 ± 24.49	10⁻¹, 2² 80 ± 24.49	10⁻², 2⁵ 50 ± 0.0	10⁻², 2⁴ 50 ± 0.0	10⁻³, 2² 50 ± 0.0	10⁻³, 2⁵ 80 ± 24.49
westEast	10⁰, 2⁵ 85 ± 22.9	10⁰, 2³ 85 ± 22.9	10⁰, 2³ 80 ± 24.49	10⁻¹, 2⁴ 50 ± 0.0	10¹, 2⁴ 50 ± 0.0	10⁻¹, 2³ 70 ± 24.49	10⁰, 2² 70 ± 24.49