eduzhai > Applied Sciences > Engineering >

Generate High Resolution Images With Generative Variational Autoencoder

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 10 pages

Abstract: In this work, we present a novel neural network to generate high resolutionimages. We replace the decoder of VAE with a discriminator while using theencoder as it is. The encoder is fed data from a normal distribution while thegenerator is fed from a gaussian distribution. The combination from both isgiven to a discriminator which tells whether the generated image is correct ornot. We evaluate our network on 3 different datasets: MNIST, LSUN and CelebAdataset. Our network beats the previous state of the art using MMD, SSIM, loglikelihood, reconstruction error, ELBO and KL divergence as the evaluationmetrics while generating much sharper images. This work is potentially veryexciting as we are able to combine the advantages of generative models andinference models in a principled bayesian manner.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×