CPGAN :  An Efficient Architecture Designing for Text-to-Image Generative Adversarial Networks Based on Canonical Polyadic Decomposition

<div>The architecture of CPGAN. The three encoding layers in the blue dotted box are the three decomposed filters obtained by decomposing the original convolution layer.</div>

Scientific Programming

fig2

Figure 2

Figure 2: CPGAN :  An Efficient Architecture Designing for Text-to-Image Generative Adversarial Networks Based on Canonical Polyadic Decomposition 