Embedded FPGA Acceleration of Brain-Like Neural Networks: Online Learning to Scalable Inference

TL;DR

This paper introduces the first embedded FPGA accelerator for Brain-Like Neural Networks, enabling low-power, scalable online learning and inference on edge devices with significant efficiency improvements.

Contribution

It presents a novel embedded FPGA implementation of BCPNN, supporting online learning and mixed precision, for practical neuromorphic edge computing.

Findings

Achieves up to 17.5x latency reduction over ARM baseline.

Realizes 94% energy savings without accuracy loss.

Demonstrates effectiveness on multiple datasets.

Abstract

Edge AI applications increasingly require models that can learn and adapt on-device with minimal energy budget. Traditional deep learning models, while powerful, are often overparameterized, energy-hungry, and dependent on cloud connectivity. Brain-Like Neural Networks (BLNNs), such as the Bayesian Confidence Propagation Neural Network (BCPNN), propose a neuromorphic alternative by mimicking cortical architecture and biologically-constrained learning. They offer sparse architectures with local learning rules and unsupervised/semi-supervised learning, making them well-suited for low-power edge intelligence. However, existing BCPNN implementations rely on GPUs or datacenter FPGAs, limiting their applicability to embedded systems. This work presents the first embedded FPGA accelerator for BCPNN on a Zynq UltraScale+ SoC using High-Level Synthesis. We implement both online learning and inference-only kernels with support for variable and mixed precision. Evaluated on MNIST, Pneumonia, and Breast Cancer datasets, our accelerator achieves up to 17.5x latency and 94% energy savings over ARM baselines, without sacrificing accuracy. This work enables practical neuromorphic computing on edge devices, bridging the gap between brain-like learning and real-world deployment.