The Echo State Network (ESN) is a type of reservoir computer that uses a recurrent neural network with a sparsely connected hidden layer. The Echo State Network diagram can be designed by using EdrawMax or EdrawMax Online. The main interest of ESN is that although its behavior is nonlinear, the only weights that are modified during training are for the synapses that connect the hidden neurons to output neurons. As the diagram suggested, the reservoir architecture creates a recurrent non-linear embedding of the input, which can then be connected to the desired output. These final weights will be trainable. It is also possible to connect the embedding to a different predictive model.
