Skip to main content

Table 5 Multi-condition (MC) classification results of different algorithms with context voting ensemble learning (-v) at the second-stage TsEL (unit: %)

From: Robust sound event classification with bilinear multi-column ELM-AE and two-stage ensemble learning

  Clean 20 dB 10 dB 0 dB Mean
DNN 96.90 96.90 93.20 80.40 91.85
ELM 93.34 ± 0.60 91.55 ± 0.57 89.22 ± 0.91 79.78 ± 1.62 88.47 ± 0.93
ELM-AE 94.23 ± 0.76 93.38 ± 1.49 91.65 ± 0.76 86.88 ± 1.60 91.54 ± 1.15
B-ELM-AE 97.38 ± 0.47 96.58 ± 0.47 95.10 ± 0.62 90.05 ± 1.35 94.78 ± 0.73
MC-ELM-AE-C 95.48 ± 0.63 94.70 ± 0.38 93.18 ± 0.91 88.73 ± 1.02 93.02 ± 0.74
MC-ELM-AE-TsEL-V 95.35 ± 0.57 94.35 ± 0.48 93.28 ± 0.82 89.00 ± 1.30 92.99 ± 0.79
MC-ELM-AE-TsEL-W 96.63 ± 0.31 95.35 ± 1.01 93.63 ± 1.02 89.50 ± 1.42 93.78 ± 0.94
B-MC-ELM-AE-C 97.13 ± 0.49 96.43 ± 0.62 94.88 ± 1.00 89.68 ± 0.97 94.53 ± 0.77
B-MC-ELM-AE-TsEL-V 97.33 ± 0.50 96.55 ± 0.60 95.38 ± 0.76 90.55 ± 0.72 94.95 ± 0.65
B-MC-ELM-AE-TsEL-W 98.03 ± 0.36 97.40 ± 0.31 96.28 ± 1.05 91.53 ± 0.84 95.81 ± 0.64