Table 3 Optimization of LSTM model parameters.
From: Deep learning time-series modeling for assessing land subsidence under reduced groundwater use
Property | Value | Property | Value |
|---|---|---|---|
Gradient threshold | 1 | Maximum epochs | 500 |
Squared gradient factor | 0.99 | Mini batch size | 10 |
Initial learn rate | 0.01 | Gradient factor | 0.9 |
Drop factor | 0.1 | Drop period | 500 |
Number of hidden layers | 350 | Solver for training network | adam |
