In our laser neural network (LNN), previously reported, a laser diode with an extended cavity is used to provide all- optical neural action. The neurons of the network are the longitudinal cavity modes of the laser. Weighted inputs are provided to these neurons via an optical matrix vector multiplier placed in the external cavity. A setup that uses a fast liquid crystal display and a loop mirror is used to expand the capabilities of the LNN to data switching applications. Up to 32 neurons can be defined having up to 12 inputs. As an example a 1:16 data decoder is demonstrated.
|Name||Proceedings of SPIE|
|Conference||1998 International Conference on Applications of Photonic Technology, ICAPT|
|Period||29/07/98 → 31/07/98|