All-Optical Neuromorphic Computing in Optical Networks of Semiconductor Lasers
Brunner, Daniel; Reitzenstein, Stephan; Fischer, Ingo
Proceedings of the IEEE International Conference on Rebooting Computing 2016, IEEE, , (2016)
Proceedings of the IEEE International Conference on Rebooting Computing 2016, IEEE, , (2016)
Networks of interconnected nodes are at the heart of every neural network concept. While neural networks have been implemented in various hardware systems, the efficient realization of such networks still represents a major challenge. We demonstrate the implementation of an all-optical network scheme based on holographic coupling and induce complex spatio-temporal transients with Gigahertz bandwidth. Our scheme illustrates the potential of such all-optical systems for future neural network implementations.
Related research projects
NeuroQNet
Neuromorphic Computing using Quantum Dot - Networks
I.P.: Ingo Fischer
The ever-increasing demand for information processing of massive data volumes at ultra-high bit rates requires novel computing concepts and their implementation in unconventional hardware. The main objective (OI) of this project is …