IEEE Computer Society Newsfeed

Subscribe RSS


« Back

Researchers Demonstrate Optoelectronic Reservoir Computing

Belgian researchers have demonstrated a new form of computing that uses feedback loops. Led by Yvan Paquot from the Université Libre de Bruxelles in Belgium, the scientists based the new system on the concept that a nonlinear feedback mechanism can take data input, process it, and generate an output. This approach is known as reservoir computing. In the system, the reservoir – which is a recurrent neural network -- is created by a large number of randomly connected nodes, each of which is a nonlinear feedback loop. Each node in this particular system is an optoelectronic device in which a voltage modulates the intensity of a beam of light. Among the tasks completed by the reservoir computer were distinguishing between sine waves and square waves, and simple word recognition. The researchers published their results at Cornell University’s arXiv print archive. (arXiv blog)(arXiv)(“Reservoir Computing Approaches to Recurrent Neural Network Training,” Mantas Lukosevicius and Herbert Jaeger at the Jacobs University, Bremen, website)

Comments
Trackback URL: