SpikySpace: A Spiking State Space Model for Energy-Efficient Time Series Forecasting
Kaiwen Tang, Jiaqi Zheng, Yuze Jin, Yupeng Qiu, Guangda Sun, Zhanglu Yan*, Weng-Fai Wong · 2026
Read SpikySpace: A Spiking State Space Model for Energy-Efficient Time Series Forecasting by Kaiwen Tang, Jiaqi Zheng, Yuze Jin, Yupeng Qiu, Guangda Sun, Zhanglu Yan*, Weng-Fai Wong (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-02411.txt.
Description
Kaiwen Tang and six co authors present SpikySpace, a forecasting model for time series that runs on spiking neural networks. The event driven networks prized for very low power use on edge devices. Forecasting traffic or industrial signals in real time needs that efficiency, yet earlier spiking forecasters relied on heavy transformer blocks with quadratic cost.
SpikySpace replaces attention with a spiking state space design and selective scanning, cutting the cost to linear in sequence length. Two cheap approximations replace costly exponential and division steps with simple bit shifts. On four multivariate benchmarks it beats the leading spiking forecaster on accuracy by up to about three percent, while cutting energy use by more than ninety six percent. The authors present it as the first fully spiking state space model.
Catalog record
| CREATOR | Kaiwen Tang, Jiaqi Zheng, Yuze Jin, Yupeng Qiu, Guangda Sun, Zhanglu Yan*, Weng-Fai Wong |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-02411