Learnable Leakage and Onset-Spiking Self-Attention in SNNs with Local Error Signals
Cong Shi, Li Wang, Haoran Gao, Min Tian, Xinqun Zhu · 2023
Read Learnable Leakage and Onset-Spiking Self-Attention in SNNs with Local Error Signals by Cong Shi, Li Wang, Haoran Gao, Min Tian, Xinqun Zhu (2023) 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/pmc-10747667.txt.
Description
Cong Shi and co-authors upgrade in 2023 spiking neural networks with learnable leak and onset attention. Spiking nets mimic brain firing and promise lean compute. Deep versions still struggle to weigh key signals across space and time in a balanced way.
Their neurons learn their own leak rate to fit modern tasks. Attention fires at the first time step to lock focus early. A new norm rule tied to the leak rate joins a local loss scheme for training. Tests on three image sets reach high accuracy in just eight time steps, a strong speed for spiking models.
Catalog record
| CREATOR | Cong Shi, Li Wang, Haoran Gao, Min Tian, Xinqun Zhu |
|---|---|
| YEAR | 2023 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 61 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |