Mikio Braun
@mikiobraun
Gregoire Montavon has a very nice preprint on how centering can lead to much better convergence in some neural networks arxiv.org/abs/1203.3783
Learning Feature Hierarchies with Centered Deep Boltzmann Machines
Deep Boltzmann machines are in principle powerful models for extracting the hierarchical structure of data. Unfortunately, attempts to train layers...
arxiv.org