Table 7 Comparison with recent state-of-the-art systems using the performance of individual systems without a score-level ensemble. The third to seventh rows list the top five best-performing systems on the DCASE2020 Task 1A challenge. The ninth to 13th rows list the top five best-performing systems on the DCASE2021 Task 1A challenge
From: Deep semantic learning for acoustic scene classification
System | Acc(%) | #Params | Compression methods |
|---|---|---|---|
Proposed method | 71.26 | 478K | - |
DCASE2020 Baseline [5] | 54.1 | 5M | - |
Suh et al. [31] | 73.7 | 13M | - |
Hu et al. [30] | 76.9 | 130M | - |
Gao et al. [32] | 71.8 | 4M | - |
Liu et al. [59] | 72.1 | 3M | - |
Koutini et al. [60] | 71.8 | 225M | - |
DCASE2021 Baseline [22] | 46.9 | 46K | Quantization |
Kim et al. [34] | 76.3 | 315K | Quantization/pruning/knowledge distillation |
Yang et al. [37] | 78.3 | 3.6K | LTH/knowledge distillation |
Koutini et al. [35] | 69.5 | 631K | Sparsify/quantization |
Heo et al. [36] | 70.5 | 65K | Knowledge distillation |
Liu et al. [61] | 68.2 | 648K | Quantization |