Respiratory sound classification on ICBHI dataset
94.55AccuracyHybrid LSTM-KAN
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Hybrid LSTM-KAN2026.01 | 94.55 | 72.92 | 69.78 | 70.33 | 94.36 | |
| CNN-LSTM hybrid2026.01 | 91.8 | 68 | 66 | 67 | 91 | |
| Standard LSTM2026.01 | 89.3 | 65 | 63 | 64 | 88.5 | |
| CNN baseline2026.01 | 86.2 | 62 | 58 | 60 | 85 | |
| Standard MLP2026.01 | 84.1 | 59 | 57 | 58 | 83 | |
| Random Forest2026.01 | 82.7 | 61 | 59 | 60 | 82 | |
| Traditional SVM approach2026.01 | 78.5 | 55 | 52 | 53.5 | 78 |