Differentiable Approximation Bridges For Training Networks Containing Non-Differentiable Functions.
Jason RamapuramRuss WebbPublished in: CoRR (2019)
Keyphrases
- loss function
- objective function
- continuous functions
- training phase
- strictly convex
- supervised learning
- network design
- social networks
- training process
- serious games
- test set
- approximation algorithms
- error bounds
- computer networks
- online learning
- heterogeneous networks
- approximation error
- training set
- echo state networks
- data sets