Optimization and applications of echo state networks with leaky- integrator neurons

@article{Jaeger2007OptimizationAA,
  title={Optimization and applications of echo state networks with leaky- integrator neurons},
  author={Herbert Jaeger and Mantas Lukosevicius and Dan Popovici and Udo Siewert},
  journal={Neural networks : the official journal of the International Neural Network Society},
  year={2007},
  volume={20 3},
  pages={
          335-52
        }
}

• Herbert Jaeger, Mantas Lukosevicius, +1 author Udo Siewert
• Published in Neural Networks 2007
• DOI:10.1016/j.neunet.2007.04.016

Standard echo state networks (ESNs) are built from simple additive units with a sigmoid activation function. Here we investigate ESNs whose reservoir units are leaky integrator units. Units of this type have individual state dynamics, which can be exploited in various ways to accommodate the network to the temporal characteristics of a learning task. We present stability conditions, introduce and investigate a stochastic gradient descent method for the optimization of the global learning… CONTINUE READING

Save to Library

Create Alert

Cite

View on PubMed