SPI-GAN: DENOISING DIFFUSION GANS WITH STRAIGHT-PATH INTERPOLATIONS

doi:doi:10.57702/srj5aqnt

SPI-GAN: DENOISING DIFFUSION GANS WITH STRAIGHT-PATH INTERPOLATIONS

Score-based generative models (SGMs) show the state-of-the-art sampling quality and diversity. However, their training/sampling complexity is notoriously high due to the highly complicated forward/reverse processes, so they are not suitable for resource-limited settings. To solving this problem, learning a simpler process is gathering much attention currently. We present an enhanced GAN-based de-noising method, called SPI-GAN, using our proposed straight-path interpolation definition.

BibTex:

@dataset{Jinsung_Jeon_and_Noseong_Park_2024,
    abstract = {Score-based generative models (SGMs) show the state-of-the-art sampling quality and diversity. However, their training/sampling complexity is notoriously high due to the highly complicated forward/reverse processes, so they are not suitable for resource-limited settings. To solving this problem, learning a simpler process is gathering much attention currently. We present an enhanced GAN-based de-noising method, called SPI-GAN, using our proposed straight-path interpolation definition.},
    author = {Jinsung Jeon and Noseong Park},
    doi = {10.57702/srj5aqnt},
    institution = {No Organization},
    keyword = {'Diffusion Models', 'Generative Adversarial Networks', 'Score-based Generative Models'},
    month = {dec},
    publisher = {TIB},
    title = {SPI-GAN: DENOISING DIFFUSION GANS WITH STRAIGHT-PATH INTERPOLATIONS},
    url = {https://service.tib.eu/ldmservice/dataset/spi-gan--denoising-diffusion-gans-with-straight-path-interpolations},
    year = {2024}
}