论文部分内容阅读
储备池计算是一种适合处理时序信号的简单高效的机器学习算法。相比在传统电子计算机上用软件实现的方式,储备池计算在光器件上的实现方式将更有利于超高速和超低功耗的信息处理。介绍了储备池计算的基本原理,从输入层、储备池和输出层三个方面介绍了储备池计算硬件实现方案的研究进展,指出了储备池计算硬件实现方案发展中存在的问题,并展望了其未来发展趋势。
Reserve pool calculation is a simple and efficient machine learning algorithm suitable for processing timing signals. Compared to the way that software is implemented on conventional electronic computers, the implementation of reserve pool computing on optical devices will be more conducive to ultra-fast and ultra-low-power information processing. The basic principle of reserve pool calculation is introduced. The research progress of reserve pool calculation hardware implementation is introduced from three aspects: input layer, reserve pool and output layer. The problems existing in the development of reserve pool calculation hardware are pointed out. Finally, Its future development trend.