A New Tandem Learning Rule for Efficient Training of Spiking Neural Network Equalizers for IM/DD Optical Transmission
Konferenz: ECOC 2024 - 50th European Conference on Optical Communication
22.09.2024-26.09.2024 in Frankfurt, Germany
Tagungsband: ITG-Fb. 317: ECOC 2024
Seiten: 4Sprache: EnglischTyp: PDF
Autoren:
Li, Shuangxu; Boecherer, Georg; Calabro, Stefano; Schaedler, Maximilian; Kong, Tianyuan
Inhalt:
A spiking neural network (SNN) equalizer for IM/DD optical transmission is trained using tandem learning, offloading the training to an artificial neural network (ANN). Optimal performance is achieved, outperforming a linear equalizer by 1 dB. The proposed learning rule is suitable for neuromorphic hardware.