Table 7 Statistics of model prediction metrics under different signal-noise ratios.
From: Application of stacked bidirectional LSTM neural networks in reservoir porosity prediction
Signal-noise ratio | Neural network model | MAE | RMSE | Predictive accuracy (%) |
|---|---|---|---|---|
10dB | RNN | 0.0109 | 0.0209 | 59.38 |
LSTM | 0.0083 | 0.0183 | 68.75 | |
Un-attention S-BiLSTM | 0.0066 | 0.0156 | 76.56 | |
S-BiLSTM | 0.0046 | 0.0136 | 82.81 | |
5dB | RNN | 0.0124 | 0.0224 | 53.75 |
LSTM | 0.0093 | 0.0192 | 64.44 | |
Un-attention S-BiLSTM | 0.0074 | 0.0169 | 73.44 | |
S-BiLSTM | 0.0054 | 0.0149 | 80.03 | |
1dB | RNN | 0.0147 | 0.0243 | 44.81 |
LSTM | 0.0125 | 0.0224 | 53.5 | |
Un-attention S-BiLSTM | 0.0089 | 0.0186 | 65.63 | |
S-BiLSTM | 0.0063 | 0.0159 | 76.12 |
