Table 3 Pseudocode for deep learning model training Process.

Input

Agricultural production dataset \(\:X\), labels \(\:Y\)

Output

Trained deep learning model

Initialization

Initialize CNN model parameters \(\:{\theta\:}_{CNN}\), LSTM model parameters \(\:{\theta\:}_{LSTM}\)

Training Process

For each training epoch:

1. Input Data

Feed \(\:X\) into the CNN network to extract spatial features

2. Combine Temporal Data

Input the CNN output together with temporal data into the LSTM network

3. Compute Loss

Compute the predicted value \(\:f\left(X\right)\) and calculate the loss \(\:L\left(\theta\:\right)\) by comparing it with the true labels \(\:Y\)

4. Update Parameters

Update the parameters \(\:{\theta\:}_{CNN}\) and \(\:{\theta\:}_{LSTM}\) using backpropagation

Output

Trained deep learning model