# VQ4SNN: Vector Quantization for Memory-Efficient FPGA Spiking Neural Networks

Source: [VQ4SNN: Vector Quantization for Memory-Efficient FPGA Spiking Neural Networks](<https://www.sota2.com/research/paper/6a3a1306206542251d7c6f00>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-24T05:00:04.550Z

Authors: Dimitrios Sekertzis, Giorgos Dimitrakopoulos

Published: 2026-06-12T15:33:36.000Z

## Abstract

Spiking Neural Networks (SNNs) offer an energy-efficient paradigm for edge AI, making them attractive for hardware acceleration. However, deploying dense SNNs on FPGAs is constrained by limited on-chip memory for synaptic weight storage. To address this bottleneck, we propose VQ4SNN, a hardware-aware architecture that reduces memory requirements through Vector Quantization (VQ)-based weight sharing. To the best of our knowledge, this is the first application of VQ to pipelined spatial-dataflow SNN accelerators on FPGAs. VQ4SNN replaces conventional weight storage with a two-level memory organization consisting of compact pointers and a shared codebook of quantized weight vectors. The proposed design integrates FPGA-aware memory mapping with analytical VQ parameter selection, enabling efficient deployment on such accelerators while preserving inference accuracy. The experimental results show a reduction of 52-61% in the total number of BRAMs compared to the state-of-the-art uncompressed FPGA SNNs without increasing overall logic utilization.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Image Classification on MNIST Large 784-392-196-10 (test)](<https://www.sota2.com/research/sota/image-classification-on-mnist-large-784-392-196-10-test.md>): Image Classification; MNIST Large 784-392-196-10 (test); Reference Accuracy: 94.38.

- [Hardware Utilization and Performance on MNIST Base 784-128-10 (test)](<https://www.sota2.com/research/sota/hardware-utilization-and-performance-on-mnist-base-784-128-10-test.md>): Image Classification; MNIST Base 784-128-10 (test); Reference Accuracy: 96.35.

- [Audio Classification on AudioMNIST 40-150-10 (test)](<https://www.sota2.com/research/sota/audio-classification-on-audiomnist-40-150-10-test.md>): Audio Classification; AudioMNIST 40-150-10 (test); Reference Accuracy: 93.12.

- [Spiking Audio Classification on SHD 700-200-20 (test)](<https://www.sota2.com/research/sota/spiking-audio-classification-on-shd-700-200-20-test.md>): Spiking Audio Classification; SHD 700-200-20 (test); Ref. Accuracy: 73.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2606.20675>)

[PDF](<https://arxiv.org/pdf/2606.20675.pdf>)
