The Numerics of GANs

@inproceedings{Mescheder2017TheNO,
  title={The Numerics of GANs},
  author={Lars M. Mescheder and Sebastian Nowozin and Andreas Geiger},
  booktitle={NIPS},
  year={2017}
}
In this paper, we analyze the numerics of common algorithms for training Generative Adversarial Networks (GANs). Using the formalism of smooth two-player games we analyze the associated gradient vector field of GAN training objectives. Our findings suggest that the convergence of current algorithms suffers due to two factors: i) presence of eigenvalues of the Jacobian of the gradient vector field with zero real-part, and ii) eigenvalues with big imaginary part. Using these findings, we design a… CONTINUE READING

Citations

Publications citing this paper.
SHOWING 1-10 OF 73 CITATIONS, ESTIMATED 42% COVERAGE

FILTER CITATIONS BY YEAR

2017
2019

CITATION STATISTICS

  • 10 Highly Influenced Citations

  • Averaged 27 Citations per year over the last 3 years

  • 81% Increase in citations per year in 2018 over 2017

References

Publications referenced by this paper.
SHOWING 1-10 OF 27 REFERENCES

Similar Papers

Loading similar papers…