Reliable Breast Cancer Diagnosis with Deep Learning: DCGAN-Driven Mammogram Synthesis and Validity Assessment

Table 2

Generator architecture.

Layer no	Layer type	Output shape	Activation
1	Dense (input)	512 × 4 × 4	—
2	Reshape	256 × 4 × 4	—
3	Conv2DTranspose	512 × 8 × 8	ReLU
4	Batch normalization	256 × 8 × 8	—
5	Conv2DTranspose	128 × 16 × 16	ReLU
6	Batch normalization	64 × 16 × 16	—
7	Conv2DTranspose	32 × 32 × 32	ReLU
8	Batch normalization	32 × 32 × 32	—
9	Conv2DTranspose	3 × 64 × 64	Tanh