Temporal dendritic heterogeneity incorporated with spiking neural networks for learning multi-timescale dynamics
文献类型：期刊论文
| 作者 | Zheng, Hanle1; Zheng, Zhong1; Hu, Rui1; Xiao, Bo1; Wu, Yujie2; Yu, Fangwen1; Liu, Xue1; Li, Guoqi3 ; Deng, Lei1
|
| 刊名 | NATURE COMMUNICATIONS
 |
| 出版日期 | 2024-01-04 |
| 卷号 | 15期号:1页码:20 |
| DOI | 10.1038/s41467-023-44614-z |
| 通讯作者 | Deng, Lei(leideng@mail.tsinghua.edu.cn) |
| 英文摘要 | It is widely believed the brain-inspired spiking neural networks have the capability of processing temporal information owing to their dynamic attributes. However, how to understand what kind of mechanisms contributing to the learning ability and exploit the rich dynamic properties of spiking neural networks to satisfactorily solve complex temporal computing tasks in practice still remains to be explored. In this article, we identify the importance of capturing the multi-timescale components, based on which a multi-compartment spiking neural model with temporal dendritic heterogeneity, is proposed. The model enables multi-timescale dynamics by automatically learning heterogeneous timing factors on different dendritic branches. Two breakthroughs are made through extensive experiments: the working mechanism of the proposed model is revealed via an elaborated temporal spiking XOR problem to analyze the temporal feature integration at different levels; comprehensive performance benefits of the model over ordinary spiking neural networks are achieved on several temporal computing benchmarks for speech recognition, visual recognition, electroencephalogram signal recognition, and robot place recognition, which shows the best-reported accuracy and model compactness, promising robustness and generalization, and high execution efficiency on neuromorphic hardware. This work moves neuromorphic computing a significant step toward real-world applications by appropriately exploiting biological observations. Brain-inspired spiking neural networks have shown their capability for effective learning, however current models may not consider realistic heterogeneities present in the brain. The authors propose a neuron model with temporal dendritic heterogeneity for improved neuromorphic computing applications. |
| WOS关键词 | EMOTION RECOGNITION ; DIVERSITY ; NEURONS ; CLASSIFICATION ; PLASTICITY ; ACCURATE |
| 资助项目 | STI[2021ZD0200300] ; National Natural Science Foundation of China[62276151] ; National Natural Science Foundation of China[62106119] ; National Natural Science Foundation of China[62236009] ; National Natural Science Foundation of China[U22A20103] ; National Science Foundation for Distinguished Young Scholars[62325603] ; Chinese Institute for Brain Research |
| WOS研究方向 | Science & Technology - Other Topics |
| 语种 | 英语 |
| 出版者 | NATURE PORTFOLIO |
| WOS记录号 | WOS:001136901800010 |
| 资助机构 | STI ; National Natural Science Foundation of China ; National Science Foundation for Distinguished Young Scholars ; Chinese Institute for Brain Research |
| 源URL | [http://ir.ia.ac.cn/handle/173211/54804]  |
| 专题 | 脑图谱与类脑智能实验室 |
| 通讯作者 | Deng, Lei |
| 作者单位 | 1.Tsinghua Univ, Ctr Brain Inspired Comp Res CBICR, Dept Precis Instrument, Beijing, Peoples R China 2.Graz Univ Technol, Inst Theoret Comp Sci, Graz, Austria 3.Chinese Acad Sci, Inst Automat, Beijing, Peoples R China |
| 推荐引用方式 GB/T 7714 | Zheng, Hanle,Zheng, Zhong,Hu, Rui,et al. Temporal dendritic heterogeneity incorporated with spiking neural networks for learning multi-timescale dynamics[J]. NATURE COMMUNICATIONS,2024,15(1):20. |
| APA | Zheng, Hanle.,Zheng, Zhong.,Hu, Rui.,Xiao, Bo.,Wu, Yujie.,...&Deng, Lei.(2024).Temporal dendritic heterogeneity incorporated with spiking neural networks for learning multi-timescale dynamics.NATURE COMMUNICATIONS,15(1),20. |
| MLA | Zheng, Hanle,et al."Temporal dendritic heterogeneity incorporated with spiking neural networks for learning multi-timescale dynamics".NATURE COMMUNICATIONS 15.1(2024):20. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。