Investigation On The Parallel Processing Ability Of A Reservoir Computing System Based On A VCSEL Under Double Optical Feedback

In today's age of big data,along with the earth-shaking change of lifestyle,the information and data faced by people also present an explosive growth.Under this background,researchers are devoting themselves to seeking for fast,efficient and precise means of processing information and data.Based on the functional mode of human brain processing external input information,artificial neural networks?ANNs?present high accuracy and high efficiency for processing some highly complex tasks such as speech recognition,handwriting recognition,facial recognition and chaotic time series prediction,and therefore show an excellent existence in the field of artificial intelligence.As a novel computing algorithm based on recurrent neural networks?RNNs?,reservoir computing?RC?has been attracting much attention of researchers due to its relatively easy training procedure.At present,in order to save the cost and make it more practical for application,many research reports mainly focus on the RC system based on single nonlinear node with a delay feedback loop?named as delay-based RC?.Among these reports,the RC systems utilizing a semiconductor laser?SL?with optical feedback as the nonlinear node are capable of processing information faster due to the high relaxation oscillation frequency,thus attracting people's interest dramatically.Up to now,many reports only involve the RC system for processing single task.However,the RC systems capable of parallel information processing may possess larger application prospects and deserve further investigation.Therefore,based on the polarization characteristics of the output light emitted by a vertical-cavity surface-emitting laser?VCSEL?,in this paper,a scheme of RC system for simultaneously processing two different tasks is proposed and its parallel processing performance is also investigated.For this scheme,a VCSEL simultaneously subject to double optical feedback and optical injection is utilized as the nonlinear node of the RC system,and a Santa-Fe time series prediction task and a wave classification task are processed based on the dynamical responses of X polarization component?X-PC?and Y polarization component?Y-PC?in the VCSEL,respectively.Considering that two different feedback frames?polarization-preserved optical feedback?PP-OF?or polarization-rotated optical feedback?PR-OF??may be adopted in two feedback loops,the parallel processing performances of RC system under four double feedback combination cases are numerically analyzed.The simulated results show that the parallel processing ability of the proposed RC system depends on the feedback frames adopted in two loops.After comprehensively evaluating the parallel processing performances of the two tasks under different feedback combinations,the best parallel processing performance can be achieved by adopting PP-OFs in both two feedback loops.Under some optimized operation parameters,this proposed RC system can realize the lowest prediction error of 0.0289 and the lowest signal classification error of2.78×10^-5.