GloVe-CNN-BiLSTM Model for Sentiment Analysis on Text Reviews

Table 2

Parameters’ setting of GloVe-CNN-BiLSTM.

Parameters	Value
Word vector dimension	300
Convolution layer filters	50
Number of convolution kernels	256
Convolution layer kernel size	2
Convolution layer padding	Valid
Convolution layer activation function	ReLU
Pooling layer pool size	1
Pooling layer activation function	ReLU
Number of hidden units in BiLSTM layer	64
Dropout rate	0.7, 0.8
Loss	categorical_crossentropy
Optimizer	Adam
Learning rate	0.0001