Parameter and Data-Efficient Spectral StyleDCGAN
This work addresses the challenge of resource-efficient image generation for applications with constrained computational or data availability, though it is incremental in improving efficiency.
The paper tackles the problem of generating high-fidelity face images with limited parameters and training data, achieving results at 64x64 resolution using only 6.574 million parameters and 4,739 dog faces from the AFHQ dataset.
We present a simple, highly parameter, and data-efficient adversarial network for unconditional face generation. Our method: Spectral Style-DCGAN or SSD utilizes only 6.574 million parameters and 4739 dog faces from the Animal Faces HQ (AFHQ) dataset as training samples while preserving fidelity at low resolutions up to 64x64. Code available at https://github.com/Aryan-Garg/StyleDCGAN.