• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International conference on artificial neural networks >Time Series Processing with VCSEL-Based Reservoir Computer
【24h】

Time Series Processing with VCSEL-Based Reservoir Computer

机译：基于VCSEL的水库计算机进行时间序列处理

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Reservoir computing architectures offer important benefits for the implementation of a neural network in a physical medium, as the weighted interconnections between the internal nodes are random and fixed. Experimental results on a time-delay photonic reservoir computer based on directly modulated Vertical Cavity Surface Emitting Lasers and multi-mode fiber couplers are presented. The neuron is made of photodiode, non-linear amplifier and laser chips. The NARMA10 chaotic time-series task is performed with a configuration having 25 virtual nodes operating at 1 GS/s. Experimental and simulated error ranges are in good agreement, which is promising for an expansion to a more elaborate system. The potential of this scheme for the realization of a photonic-reservoir cluster device operating at very high speed with low power and a small footprint with a large number of interacting physical and virtual neurons is discussed.
机译：储层计算体系结构为在物理介质中实现神经网络提供了重要的好处，因为内部节点之间的加权互连是随机且固定的。提出了基于直接调制垂直腔表面发射激光器和多模光纤耦合器的时滞光子储层计算机的实验结果。神经元由光电二极管，非线性放大器和激光芯片组成。 NARMA10混沌时间序列任务是使用具有25个以1 GS / s的速度运行的虚拟节点的配置执行的。实验和模拟误差范围非常吻合，这有望扩展到更复杂的系统。讨论了该方案对于实现以非常高的速度以低功率和小占用空间以及大量相互作用的物理和虚拟神经元运行的光子-储层集群设备的潜力。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号