Federated Few-Shot Learning on Neuromorphic Hardware: An Empirical Study Across Physical Edge Nodes

TL;DR

This empirical study explores federated learning on neuromorphic hardware, demonstrating that neuron-level weight exchange strategies preserve accuracy and that feature quality significantly impacts federated learning success across physical edge nodes.

Contribution

The paper provides the first empirical evaluation of federated learning strategies on neuromorphic hardware, highlighting effective weight exchange methods and the importance of feature quality.

Findings

Neuron-level concatenation (FedUnion) preserves accuracy.

Element-wise weight averaging (FedAvg) destroys accuracy.

Scaling feature dimensionality improves federated accuracy.

Abstract

Federated learning on neuromorphic hardware remains unexplored because on-chip spike-timing-dependent plasticity (STDP) produces binary weight updates rather than the floating-point gradients assumed by standard algorithms. We build a two-node federated system with BrainChip Akida AKD1000 processors and run approximately 1,580 experimental trials across seven analysis phases. Of four weight-exchange strategies tested, neuron-level concatenation (FedUnion) consistently preserves accuracy while element-wise weight averaging (FedAvg) destroys it (p = 0.002). Domain-adaptive fine-tuning of the upstream feature extractor accounts for most of the accuracy gains, confirming feature quality as the dominant factor. Scaling feature dimensionality from 64 to 256 yields 77.0% best-strategy federated accuracy (n=30, p < 0.001). Two independent asymmetries (wider features help federation more than individual learning, while binarization hurts federation more) point to a shared prototype complementarity mechanism: cross-node transfer scales with the distinctiveness of neuron prototypes.