Expressivity of Hidden Markov Chains vs. Recurrent Neural Networks From a System Theoretic Viewpoint.
François DesbouvriesYohan PetetinAchille SalaünPublished in: IEEE Trans. Signal Process. (2023)
Keyphrases
- recurrent neural networks
- markov chain
- viewpoint
- steady state
- neural network
- transition probabilities
- recurrent networks
- finite state
- feed forward
- random walk
- monte carlo
- stationary distribution
- markov process
- stochastic process
- markov model
- markov processes
- reservoir computing
- echo state networks
- cascade correlation
- state space
- artificial neural networks
- monte carlo method
- transition matrix
- nonlinear dynamic systems
- neural model
- search space
- probabilistic automata
- expert systems