NeuroMorse
收藏资源简介:
NeuroMorse数据集是由詹姆斯库克大学科学和工程学院等机构创建的,专为神经形态学习系统基准测试设计的时态结构化数据集。该数据集将英语中最常用的50个单词转换为莫尔斯电码的尖峰序列,尽管只使用两个输入尖峰通道,但通过时态模式编码了复杂信息。数据集在多个时间尺度上具有特征层次结构,能够测试神经形态算法分解输入模式成为空间和时间层次结构的能力。该数据集旨在解决神经形态计算中的时态特征识别问题。
The NeuroMorse dataset is a temporally structured dataset developed by the College of Science and Engineering at James Cook University and other relevant institutions, specifically designed for benchmarking neuromorphic learning systems. It converts the 50 most frequently used English words into spike sequences corresponding to Morse code, encoding complex information through temporal patterns while utilizing only two input spike channels. The dataset exhibits hierarchical features across multiple timescales, enabling the assessment of neuromorphic algorithms' ability to decompose input patterns into spatial and temporal hierarchies. This dataset aims to address the challenge of temporal feature recognition in the field of neuromorphic computing.




