EqSpike: Spike-driven Equilibrium Propagation for Neuromorphic Implementations
This addresses the problem of energy-efficient, locally constrained learning for neuromorphic hardware, offering a novel spike-driven approach with incremental improvements over existing methods.
The paper tackled the challenge of developing spike-based learning algorithms for neuromorphic systems by introducing EqSpike, a spiking neural network algorithm using Equilibrium Propagation, achieving 97.6% accuracy on MNIST and potential energy reductions of three orders of magnitude for inference and two orders for training compared to GPUs.
Finding spike-based learning algorithms that can be implemented within the local constraints of neuromorphic systems, while achieving high accuracy, remains a formidable challenge. Equilibrium Propagation is a promising alternative to backpropagation as it only involves local computations, but hardware-oriented studies have so far focused on rate-based networks. In this work, we develop a spiking neural network algorithm called EqSpike, compatible with neuromorphic systems, which learns by Equilibrium Propagation. Through simulations, we obtain a test recognition accuracy of 97.6% on MNIST, similar to rate-based Equilibrium Propagation, and comparing favourably to alternative learning techniques for spiking neural networks. We show that EqSpike implemented in silicon neuromorphic technology could reduce the energy consumption of inference and training respectively by three orders and two orders of magnitude compared to GPUs. Finally, we also show that during learning, EqSpike weight updates exhibit a form of Spike Timing Dependent Plasticity, highlighting a possible connection with biology.