Table 4 The parameters of the BiLSTM-CNN-attention structure.

From: Attention-enhanced and integrated deep learning approach for fishing vessel classification based on multiple features

Parameter

Value

Epoches

100

Learning rate

0.001

Batch size

64

Activation function

ReLU

Pooling

Average2D

Optimizer

Adam

Output activation function

Softmax

Loss function

categorical_crossentropy

Back to article page