Variance-Preserving Initialization Schemes Improve Deep Network Training: But Which Variance is Preserved?
Kyle LutherH. Sebastian SeungPublished in: CoRR (2019)
Keyphrases
- correlation coefficient
- low variance
- prediction error
- supervised learning
- network traffic
- neural network structure
- data sets
- allocation scheme
- network topologies
- radial basis function network
- network architecture
- complex networks
- covariance matrix
- network structure
- training examples
- model selection
- peer to peer
- k means