[Retracted] Software Systems Security Vulnerabilities Management by Exploring the Capabilities of Language Models Using NLP

Table 2

DCNN model construct.

Layer	Configuration
Embedding	Vocabulary size, embedding dimension
Convolutional 1D bigram	Kernel size = 2; padding = valid; activation = ReLU
Convolutional 1D trigram	Kernel size = 3; padding = valid; activation = ReLU
Convolutional 1D fourgram	Kernel size = 4; padding = valid; activation = ReLU
Pooling	GlobalMaxPool1D
Dense	Activation = ReLU
Dropout	0.2
Last dense (for 2 classes)	Units = 1; activation = sigmoid
Last dense (for more than 2 classes)	Units = number of classes; activation = softmax
Loss (for 2 classes)	Binary cross-entropy
Loss (for multiclass)	Sparse categorical cross-entropy