Expressivity of Hidden Markov Chains vs. Recurrent Neural Networks from a system theoretic viewpoint.
François DesbouvriesYohan PetetinAchille SalaünPublished in: CoRR (2022)
Keyphrases
- recurrent neural networks
- markov chain
- viewpoint
- steady state
- feed forward
- finite state
- transition probabilities
- monte carlo
- neural network
- state space
- random walk
- markov model
- recurrent networks
- reservoir computing
- artificial neural networks
- markov process
- markov processes
- monte carlo method
- stationary distribution
- neural model
- stochastic process
- echo state networks
- assemble to order systems
- cascade correlation
- transition matrix
- probabilistic automata
- nonlinear dynamic systems
- decision making
- random numbers