[Retracted] Component-Based Software Testing Method Based on Deep Adversarial Network

Table 1

Analysis of deep adversarial network models.

Deep adversarial networks	Advantage	Shortcoming	Application
GAN model	The director of the two sets	It is easy to make mistakes in the game and cause the system to crash	Data augmentation
Discriminative model	Easy to learn, average performance	Just calculate the interface and solve the problem roughly	Creative arts, stylized
Generative model	Fast convergence, learning distributions, estimating variables	Learning complex	Image generation