Negatively Correlated Echo State Networks

Echo State Network (ESN) is a special type of recurrent neu- ral network with fixed random recurrent part (reservoir) and a trainable reservoir-to-output readout mapping (typically obtained by linear regres- sion). In this work we utilise an ensemble of ESNs with diverse reservoirs whose collective read-out is obtained through Negative Correlation Learn- ing (NCL) of ensemble of Multi-Layer Perceptrons (MLP), where each individual MPL realises the readout from a single ESN. Experimental re- sults on three data sets confirm that, compared with both single ESN and flat ensembles of ESNs, NCL based ESN ensembles achieve better gener- alisation performance.