Predicting Slurry Pressure Balance with a Long Short-Term Memory Recurrent Neural Network in Difficult Ground Condition

Figure 4

(a) Recurrent neural network and its unfolding in time, where x, h, and O are, respectively, the input layer, hidden layer, and output layer, while U, W, and V are the corresponding weight matrixes; (b) in contrast, the LSTM’s hidden layer has one more component, c, the cell state.

(a)

(b)