基于光纤中拉曼散射原理实现多阈值神经网络

The design of Multi-thresholded NN based on Stimulated Raman Scattering

本文根据多路光波系统中发生的受激喇曼散射理论实现了能实现异或运算的多阈值的神经元。神经元的阈值通过本地光波信号功率、信道间隔、频率的调整来实现。信号输入通过不同的功率来表示,并通过单个多阈值神经元实现全加运算的多阈值神经网络。并通过理论分析了其可行性。

The paper introduces the theory the design of Multi-thresholded XOR neuron based on Stimulated Raman Scattering in density wavelength division multiplexed transmission systems. The thresholded of neuron is adjusted by optical signal power or channel interval or signal frequency . Different power show different sign Multi-thresholded neuron constitute NN. And the theory is validated.