A simple Python implementation of Bayes by Backprop. The original paper is "Weight Uncertainty in Neural Networks", Charles Blundell, Julien Cornebise, Koray Kavukcuoglu, and Daan Wierstra: https://arxiv.org/abs/1505.05424
- Python (2.7.6)
- Chainer (1.16.0)
- six (1.10.0)
- numpy (1.11.1)
- scikit-learn (0.17.1)
at least worked well on Ubuntu 14.04
python bayes_by_backprop.py
Still incomplete (Nov, 2016) If you find bugs, I hope you could repote them.