Skip to content

Latest commit

 

History

History
6 lines (5 loc) · 280 Bytes

README.md

File metadata and controls

6 lines (5 loc) · 280 Bytes

keras-bda

Train the three sub-networks in the paper "A Bayesian Data Augmentation Approach for Learning Deep Models" (NIPS 2017). The baseline classifiers are Lenet5, ResNet18, ResNetpa on the MNIST dataset . The code was adapted from: https://github.com/lukedeo/keras-acgan