Adversarial Training with Voronoi Constraints

@article{Khoury2019AdversarialTW,
  title={Adversarial Training with Voronoi Constraints},
  author={Marc Khoury and Dylan Hadfield-Menell},
  journal={ArXiv},
  year={2019},
  volume={abs/1905.01019}
}
Adversarial examples are a pervasive phenomenon of machine learning models where seemingly imperceptible perturbations to the input lead to misclassifications for otherwise statistically accurate models. We propose a geometric framework, drawing on tools from the manifold reconstruction literature, to analyze the high-dimensional geometry of adversarial examples. In particular, we highlight the importance of codimension: for low-dimensional data manifolds embedded in high-dimensional space… CONTINUE READING

Citations

Publications citing this paper.

Adversarial Examples for Non-Parametric Methods: Attacks, Defenses and Large Sample Limits

Yao-Yuan Yang, Cyrus Rashtchian, Yizhen Wang, Kamalika Chaudhuri
  • ArXiv
  • 2019
VIEW 1 EXCERPT
CITES BACKGROUND

References

Publications referenced by this paper.
SHOWING 1-10 OF 39 REFERENCES

Fixed Points of the Restricted Delaunay Triangulation Operator

  • Symposium on Computational Geometry
  • 2016
VIEW 6 EXCERPTS
HIGHLY INFLUENTIAL

Surface Reconstruction by Voronoi Filtering

  • Discrete & Computational Geometry
  • 1999
VIEW 7 EXCERPTS
HIGHLY INFLUENTIAL

Manifold Reconstruction Using Tangential Delaunay Complexes

  • Discrete & Computational Geometry
  • 2014
VIEW 3 EXCERPTS
HIGHLY INFLUENTIAL

Similar Papers

Loading similar papers…