Corpus ID: 59608814

Radial and Directional Posteriors for Bayesian Neural Networks

@article{Oh2019RadialAD,
  title={Radial and Directional Posteriors for Bayesian Neural Networks},
  author={Changyong Oh and K. Adamczewski and Mijung Park},
  journal={ArXiv},
  year={2019},
  volume={abs/1902.02603}
}
  • Changyong Oh, K. Adamczewski, Mijung Park
  • Published 2019
  • Computer Science, Mathematics
  • ArXiv
  • We propose a new variational family for Bayesian neural networks. We decompose the variational posterior into two components, where the radial component captures the strength of each neuron in terms of its magnitude; while the directional component captures the statistical dependencies among the weight parameters. The dependencies learned via the directional density provide better modeling performance compared to the widely-used Gaussian mean-field-type variational family. In addition, the… CONTINUE READING
    13 Citations

    Figures, Tables, and Topics from this paper

    Radial Bayesian Neural Networks: Robust Variational Inference In Big Models
    • 5
    • PDF
    Efficient Low Rank Gaussian Variational Inference for Neural Networks
    • 1
    • PDF
    Liberty or Depth: Deep Bayesian Neural Nets Do Not Need Complex Weight Posterior Approximations
    • 3
    • PDF
    Dirichlet Pruning for Neural Network Compression
    • PDF
    Uncertainty Quantification in Deep Learning for Safer Neuroimage Enhancement
    • 14
    • PDF
    Benchmarking Bayesian Deep Learning with Diabetic Retinopathy Diagnosis
    • 11
    • PDF

    References

    SHOWING 1-10 OF 58 REFERENCES
    Functional Variational Bayesian Neural Networks
    • 86
    • PDF
    Practical Variational Inference for Neural Networks
    • A. Graves
    • Computer Science, Mathematics
    • NIPS
    • 2011
    • 752
    • Highly Influential
    • PDF
    Noisy Natural Gradient as Variational Inference
    • 90
    • PDF
    Structured and Efficient Variational Deep Learning with Matrix Gaussian Posteriors
    • 137
    • Highly Influential
    • PDF
    Variational Dropout Sparsifies Deep Neural Networks
    • 417
    • Highly Influential
    • PDF
    Probabilistic Backpropagation for Scalable Learning of Bayesian Neural Networks
    • 488
    • Highly Influential
    • PDF
    Weight Uncertainty in Neural Networks
    • 994
    • Highly Influential
    • PDF
    Mean field variational Bayes for continuous sparse signal shrinkage: Pitfalls and remedies
    • 41
    • PDF