Synaptic Plasticity Dynamics for Deep Continuous Local Learning (DECOLLE)

TLDR

Biological neural networks share striking similarities with recurrent binary networks, yet aligning deep learning dynamics with spiking neural network plasticity is hindered by mismatches between synaptic dynamics and gradient backpropagation requirements. The study proposes using locally synthesized gradients to approximate backpropagation, thereby overcoming the mismatch between synaptic dynamics and gradient requirements. DECOLLE learns deep spatio‑temporal representations from spikes using local information, with synaptic plasticity rules derived from cost functions and dynamics via automatic differentiation. Synthetic gradients enable DECOLLE, which achieves state‑of‑the‑art performance on N‑MNIST and DvsGesture, offering continuously learning, biologically relevant, low‑power event‑based vision systems with accuracy comparable to conventional computers.

Abstract

A growing body of work underlines striking similarities between biological neural networks and recurrent, binary neural networks. A relatively smaller body of work, however, discusses similarities between learning dynamics employed in deep Artificial Neural Network and synaptic plasticity in spiking neural networks. The challenge preventing this is largely caused by the discrepancy between the dynamical properties of synaptic plasticity and the requirements for gradient backpropagation. Learning algorithms that approximate gradient backpropagation using locally synthesized gradients can overcome this challenge. Here, we show that synthetic gradients enable the derivation of Deep Continuous Local Learning (DECOLLE) in spiking neural networks. DECOLLE is capable of learning deep spatio-temporal representations from spikes relying solely on local information. Synaptic plasticity rules are derived systematically from user-defined cost functions and neural dynamics by leveraging existing autodifferentiation methods of machine learning frameworks. We benchmark our approach on the event-based neuromorphic dataset N-MNIST and DvsGesture, on which DECOLLE performs comparably to the state-of-the-art. DECOLLE networks provide continuously learning machines that are relevant to biology and supportive of event-based, low-power computer vision architectures matching the accuracies of conventional computers on tasks where temporal precision and speed are essential.