RNNs Evolving on an Equilibrium Manifold: A Panacea for Vanishing and Exploding Gradients?
Anil KagZiming ZhangVenkatesh SaligramaPublished in: CoRR (2019)
Keyphrases
- recurrent neural networks
- manifold learning
- fuzzy neural network
- feedforward neural networks
- nash equilibrium
- game theory
- multiscale
- games with incomplete information
- echo state networks
- parameter space
- euclidean space
- riemannian manifolds
- fisher information
- mixed strategy
- input output
- low dimensional
- lower dimensional
- high dimensional
- feature space
- tangent space
- genetic algorithm