Organic - Reservoir Computing (EU FP7)

See the Organic Website Here:  http://www.reservoir-computing.org/organic

The fundamental principle of RC, which distinguishes it from other views on recurrent neural networks, can be summarized as follows (see also thisScholarpedia article):

·                        use a large, random RNN as an excitable medium - called a reservoir in this context -, such that when driven by input signals, each unit in the RNN creates its own nonlinear transform of the input;

·                        output signals are read out from the excited RNN by some readout mechanism, typically a simple linear combination of the reservoir signals;

·                        outputs can be trained in a supervised way, typically by linear regression of the teacher output on the tapped reservoir signals.

Reservoir computing, as a recently coined term, subsumes a number of independently found instantiations of this fundamental idea:

·                        Temporal Recurrent Neural Network (Dominey 1995)

·                        Liquid State Machines (Natschläger, Maass and Markram 2002)

·                        Echo State Networks (Jaeger 2001)

·                        Decorrelation-Backpropagation Learning (Steil 2004)

Back

Print