Deep Tree Photonic Reservoir Computing with Parallel Architecture
Ronghua Zhang, Liyue Zhang, Song-Sui Li, Wei Pan, Lianshan Yan, Bin Luo, Xihua Zou, Ling Peng · 2024
In this paper, we investigate the parallel deep tree reservoir computing structure based on semiconductor lasers. The reservoir computing is reconstructed with the combination of deep reservoir computing and tree structure. The aim of this architecture is to enhance the information processing speed while maintaining robust computational performance by leveraging the strengths of both deep reservoirs and parallel reservoirs. We sub-jected our model to rigorous testing with the Santa Fe task and achieved a commendable performance, attaining an error rate as low as$1.28\times 10^{-3}$. Concurrently, we conducted comparative studies with the traditional deep reservoir and observed that the parallel architecture can effectively transcend the limitations imposed by the number of layers typically associated with deep reservoirs. We elucidate that the parallel deep tree reservoir structure can significantly enhances the computing performance both in terms of both in terms of efficiency and effectiveness.