Table 3 DNN regression performance for different numbers of hidden layers and nodes per layer
From: Robust energy disaggregation using appliance-specific temporal contextual information
Layers/nodes | 4 | 8 | 16 | 32 | 64 | 128 |
|---|---|---|---|---|---|---|
1 layer | 80.40% | 87.50% | 87.90% | 83.70% | 86.40% | 81.70% |
2 layers | 70.10% | 86.40% | 86.90% | 87.50% | 82.70% | 83.60% |
3 layers | 80.40% | 86.70% | 87.90% | 88.70% | 88.40% | 84.20% |
4 layers | 75.40% | 87.95% | 87.02% | 87.15% | 85.32% | x |