@misc{indiciae359d75fb3caa,
  title = {Unsupervised pre-training helps to conserve views from input distribution},
  author = {Nicolas Pinchaud},
  year = {2019},
  url = {https://arxiv.org/abs/1905.12889},
  note = {Source identifier: 1905.12889}
}