SuperSpike: Supervised Learning in Multilayer Spiking Neural Networks

By Friedemann Zenke, Surya Ganguli. Neural Computation, 2017

696 citations recorded by the source corpus.

A vast majority of computation in the brain is performed by spiking neural networks. Despite the ubiquity of such spiking, we currently lack an understanding of how biological spiking neural circuits learn and compute in vivo, as well as how we can instantiate such capabilities in artificial spiking circuits in silico. Here we revisit the problem of supervised learning in temporally coding multilayer spiking neural networks. First, by using a surrogate gradient approach, we derive SuperSpike, a nonlinear voltage-based three-factor learning rule capable of training multilayer networks of deterministic integrate-and-fire neurons to perform nonlinear computations on spatiotemporal spike patterns. Second, inspired by recent results on feedback alignment, we compare the performance of our learning rule under different credit assignment strategies for propagating output errors to hidden units. Specifically, we test uniform, symmetric, and random feedback, finding that simpler tasks can be solved with any type of feedback, while more complex tasks require symmetric feedback. In summary, our results open the door to obtaining a better scientific understanding of learning and computation in spiking neural networks by advancing our ability to train them to solve nonlinear problems involving transformations between different spatiotemporal spike time patterns.

Subfields: Artificial Intelligence, Cognitive Neuroscience, Electrical and Electronic Engineering

View Paper: https://www.semanticscholar.org/paper/d137505bb9617d990cd386943b146bd3294ba26d

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Cognitive Neuroscience

In Electrical and Electronic Engineering

Citations in the corpus

Theories of Error Back-Propagation in the Brain (2019, background): “…(e.g., handwritten digit classi ﬁ cation) [12 – 14, The criticism of weight symmetry has been addressed by demonstrating that even if the errors in arti ﬁ cial neural networks are back-propagated by random connections, good performance in classi ﬁ cation tasks can still be achieved [21,23 – 27].”
Deep Learning in Spiking Neural Networks (2018, background): “(cf. [57], Section III-A of this review). These approaches, however, are still useful from both engineering and scientiﬁc standpoints. Progress on the second issue has recently been made by [103] and [104]. It was shown in [103] that for some tasks, backpropagation could still perform well if random”
Enabling Deep Spiking Neural Networks with Hybrid Conversion and Spike Timing Dependent Backpropagation (2020, methodology): “Zenke & Ganguli (2018) derived a surrogate gradient based method on the membrane potential of a spiking neuron at a single time step only.”
Hebbian Learning based Orthogonal Projection for Continual Learning of Spiking Neural Networks (2024, background): “Some works introduce eligibility traces to improve BPTT for online training through time (Zenke & Ganguli, 2018; Bellec et al., 2020; Bohnstingl et al., 2022; Xiao et al., 2022), in which OTTT is an efficient and scalable approach with more theoretical guarantees.”
Artificial Neural Networks for Neuroscientists: A Primer (2020, methodology): “However, several recent methods have been proposed to train spiking networks with gradientbased techniques (Courbariaux et al., 2016; Bellec et al., 2018; Zenke and Ganguli, 2018; Nicola and Clopath, 2017; Huh and Sejnowski, 2018).”
Surrogate Gradient Learning in Spiking Neural Networks: Bringing the Power of Gradient-based optimization to spiking neural networks (2019, methodology): “S(0)[0] S(0)[1] S(0)[2] I(1)[0] I(1)[1] I(1)[2] U(1)[0] U(1)[1] U(1)[2] W(2) W(2)”
A solution to the learning dilemma for recurrent networks of spiking neurons (2019, methodology): “In contrast to e-prop , none of the papers above (Zenke and Ganguli, 2018; Murray, 2019; Roth et al., 2019) derived a theory or a deﬁnition of eligibility traces that can be applied to neuron models with a non-trivial internal dynamics, such as adaptive neurons or LSTM units, that appear to be…”
Spiking Neural Networks and Their Applications: A Review (2022, methodology): “Reference [48] proposed an improved method of SpikeProp called SuperSpike that utilizes the derivative of the membrane potential instead of the spike, which allows training a model with an absence of spike occurrence.”
Opportunities for neuromorphic computing algorithms and applications (2022, methodology): “To leverage the temporal dimension of SNN, some learning algorithms aim to generate single or multiple spikes at desired times, which have been applied in classification tasks 79–83 .”
SLAYER: Spike Layer Error Reassignment in Time (2018, background): “(15) The negative portion of fast sigmoid function, used in [14], is also a suitable candidate for ρ(u(t)−θ).”
A review of learning in biologically plausible spiking neural networks (2019, positive): “However, these DSNNs are close to classical neural networks which work based on continuous values, and they neglect the information carried by individual spikes that can lead to a fast computation (Zenke & Ganguli, 2018).”