Pre-training on physical dynamics enhances indirect sensing in recurrent neural networks.
Mitsuhiro NishidaRyo SakuraiYasumichi WakaoKohei NakajimaPublished in: MHS (2022)
Keyphrases
- recurrent neural networks
- recurrent networks
- echo state networks
- feedforward neural networks
- feed forward
- neural network
- complex valued
- artificial neural networks
- reservoir computing
- cascade correlation
- hidden layer
- feed forward neural networks
- long short term memory
- neural model
- hebbian learning
- biologically inspired
- nonlinear dynamic systems
- back propagation
- physical world
- training algorithm
- dynamic model
- sensor networks
- knowledge base
- learning algorithm