Corpus ID: 224722235

Statistical guarantees for generative models without domination.

@article{Schreuder2020StatisticalGF,
  title={Statistical guarantees for generative models without domination.},
  author={Nicolas Schreuder and Victor-Emmanuel Brunel and A. Dalalyan},
  journal={arXiv: Statistics Theory},
  year={2020}
}
  • Nicolas Schreuder, Victor-Emmanuel Brunel, A. Dalalyan
  • Published 2020
  • Mathematics
  • arXiv: Statistics Theory
  • In this paper, we introduce a convenient framework for studying (adversarial) generative models from a statistical perspective. It consists in modeling the generative device as a smooth transformation of the unit hypercube of a dimension that is much smaller than that of the ambient space and measuring the quality of the generative model by means of an integral probability metric. In the particular case of integral probability metric defined through a smoothness class, we establish a risk bound… CONTINUE READING

    Figures from this paper

    References

    SHOWING 1-10 OF 42 REFERENCES
    Generalization Properties of Optimal Transport GANs with Latent Distribution Learning
    • 2
    • Highly Influential
    • PDF
    On How Well Generative Adversarial Networks Learn Densities: Nonparametric and Parametric Results
    • 34
    • Highly Influential
    • PDF
    Some Theoretical Properties of GANs
    • 16
    • Highly Influential
    • PDF
    Statistical Guarantees of Generative Adversarial Networks for Distribution Estimation
    • 5
    • PDF
    Generative Adversarial Nets
    • 20,123
    • Highly Influential
    • PDF
    Learning in Implicit Generative Models
    • 235
    • PDF
    Learning Generative Models with Sinkhorn Divergences
    • 214
    • PDF
    Nonparametric Density Estimation with Adversarial Losses
    • 7
    • PDF
    Wasserstein Generative Adversarial Networks
    • 2,251
    • PDF