Corpus ID: 196831582

On the "steerability" of generative adversarial networks

@article{Jahanian2020OnT,
  title={On the "steerability" of generative adversarial networks},
  author={A. Jahanian and L. Chai and Phillip Isola},
  journal={ArXiv},
  year={2020},
  volume={abs/1907.07171}
}
  • A. Jahanian, L. Chai, Phillip Isola
  • Published 2020
  • Computer Science
  • ArXiv
  • An open secret in contemporary machine learning is that many models work beautifully on standard benchmarks but fail to generalize outside the lab. [...] Key Method In particular, we study their ability to fit simple transformations such as camera movements and color changes. We find that the models reflect the biases of the datasets on which they are trained (e.g., centered objects), but that they also exhibit some capacity for generalization: by "steering" in latent space, we can shift the distribution while…Expand Abstract
    Generative Hierarchical Features from Synthesizing Images
    InterFaceGAN: Interpreting the Disentangled Face Representation Learned by GANs
    2
    Interpreting the Latent Space of GANs for Semantic Face Editing
    63
    In-Domain GAN Inversion for Real Image Editing
    8
    Image Processing Using Multi-Code GAN Prior
    9

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 28 REFERENCES
    GAN Dissection: Visualizing and Understanding Generative Adversarial Networks
    122
    Large Scale GAN Training for High Fidelity Natural Image Synthesis
    1092
    A Style-Based Generator Architecture for Generative Adversarial Networks
    1033
    Generative Adversarial Nets
    16768
    Interpreting the Latent Space of GANs for Semantic Face Editing
    63
    Generative Visual Manipulation on the Natural Image Manifold
    644
    GANalyze: Toward Visual Definitions of Cognitive Image Properties
    32