Corpus ID: 211818022

Fast Predictive Uncertainty for Classification with Bayesian Deep Networks

@article{Hobbhahn2020FastPU,
  title={Fast Predictive Uncertainty for Classification with Bayesian Deep Networks},
  author={Marius Hobbhahn and Agustinus Kristiadi and Philipp Hennig},
  journal={ArXiv},
  year={2020},
  volume={abs/2003.01227}
}
  • Marius Hobbhahn, Agustinus Kristiadi, Philipp Hennig
  • Published 2020
  • Mathematics, Computer Science
  • ArXiv
  • In Bayesian Deep Learning, distributions over the output of classification neural networks are approximated by first constructing a Gaussian distribution over the weights, then sampling from it to receive a distribution over the categorical output distribution. This is costly. We reconsider old work to construct a Dirichlet approximation of this output distribution, which yields an analytic map between Gaussian distributions in logit space and Dirichlet distributions (the conjugate prior to the… CONTINUE READING

    Figures, Tables, and Topics from this paper.

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 38 REFERENCES

    Predictive Uncertainty Estimation via Prior Networks

    VIEW 4 EXCERPTS
    HIGHLY INFLUENTIAL