Spiker-LL: An Energy-Efficient FPGA Accelerator Enabling Adaptive Local Learning in Spiking Neural Networks
- Post by: admin
- May 21, 2026
- No Comment
Spiker is starting to learn! ðŸ§
Our new paper “SPIKER-LL: An Energy-Efficient FPGA Accelerator Enabling Adaptive Local Learning in Spiking Neural Networks” is now on arXiv.
The focus is the integration of the STSF local learning rule into the Spiker+ architecture, adding on-device training capabilities with minimal hardware overhead while remaining DSP-free and sub-millisecond.
We will be presenting this work at ISVLSI 2026 in Kolkata and as a poster at DAC 2026 in Long Beach — see you there!
The code will soon be available on the official Spiker+ repository.
📄 Paper: https://arxiv.org/abs/2605.18003
💻 Code: https://github.com/smilies-polito/Spiker
Thanks to my co-authors, PhD student Filippo Marostica, professors Alessandro Savino and Stefano Di Carlo and to the SMILIES Polito Research Group for their contributions to this work.
A special thanks to Alessio Carpegna, whose work gave birth to the Spiker+ architecture at the foundation of this research.
Shared by: Alessio Caviglia
View original post on LinkedIn