Table 5.

Terms used in deep learning models for the model’s optimization and prediction

Term	Definition
Activation function	The function that produces the neuron’s output
Batch	Partition of the data into different sets within a given epoch
Dropout	Removal of a fixed number of neurons during each training set for controlling overfitting
Early stopping	A strategy to control overfitting by the early stopping of the model
Epoch	Shifting the training set into different batches
Neuron	A primary entity of the DL model which learns the information and provides the output to the next layer using different activation functions
Regularization	Works as a penalty for neurons’ weight during model training