SuperSpike: Supervised learning in multi-layer spiking neural networks

Abstract

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 multi-layer spiking neural networks. First, by using a surrogate gradient approach, we derive SuperSpike, a nonlinear voltage-based three factor learning rule capable of training multi-layer 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.

References in corpus (6)

• Deep Learning in Neural Networks: An Overview
• Supervised Learning in Multilayer Spiking Neural Networks
• Learning universal computations with spikes
• Predicting non-linear dynamics by stable local learning in a recurrent spiking neural network
• Supervised Learning in Spiking Neural Networks for Precise Temporal Encoding
• Encoding Spike Patterns in Multilayer Spiking Neural Networks

Cited by in corpus (53)

• Deep Learning in Spiking Neural Networks
• Artificial neural networks for neuroscientists: A primer
• Synaptic Plasticity Dynamics for Deep Continuous Local Learning (DECOLLE)
• The Heidelberg spiking datasets for the systematic evaluation of spiking neural networks
• Spiking Neural Networks Hardware Implementations and Challenges: a Survey
• S4NN: temporal backpropagation for spiking neural networks with one spike per neuron
• MorphIC: A 65-nm 738k-Synapse/mm${}^2$ Quad-Core Binary-Weight Digital Neuromorphic Processor with Stochastic Spike-Driven Online Learning
• Adaptive Extreme Edge Computing for Wearable Devices
• SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks with at most one Spike per Neuron
• Event-Based Backpropagation can compute Exact Gradients for Spiking Neural Networks
• Bottom-up and top-down approaches for the design of neuromorphic processing systems: Tradeoffs and synergies between natural and artificial intelligence
• Federated Learning with Spiking Neural Networks
• BS4NN: Binarized Spiking Neural Networks with Temporal Coding and Learning
• Is Neuromorphic MNIST neuromorphic? Analyzing the discriminative power of neuromorphic datasets in the time domain
• Training dynamically balanced excitatory-inhibitory networks
• The Backpropagation Algorithm Implemented on Spiking Neuromorphic Hardware
• Fluctuation-driven initialization for spiking neural network training
• Stochasticity from function -- why the Bayesian brain may need no noise
• Spiking neural networks trained with backpropagation for low power neuromorphic implementation of voice activity detection
• Exploiting Noise as a Resource for Computation and Learning in Spiking Neural Networks
• Integration of Leaky-Integrate-and-Fire-Neurons in Deep Learning Architectures
• ETLP: Event-based Three-factor Local Plasticity for online learning with neuromorphic hardware
• An optimised deep spiking neural network architecture without gradients
• Loss shaping enhances exact gradient learning with Eventprop in spiking neural networks
• Differentiable graph-structured models for inverse design of lattice materials
• Embodied Synaptic Plasticity with Online Reinforcement learning
• Smooth Exact Gradient Descent Learning in Spiking Neural Networks
• Automotive Object Detection via Learning Sparse Events by Spiking Neurons
• Visual Pattern Recognition with on On-chip Learning: towards a Fully Neuromorphic Approach
• Gradient-descent hardware-aware training and deployment for mixed-signal Neuromorphic processors
• Paired Competing Neurons Improving STDP Supervised Local Learning In Spiking Neural Networks
• A Homomorphic Encryption Framework for Privacy-Preserving Spiking Neural Networks
• Target spiking patterns enable efficient and biologically plausible learning for complex temporal tasks
• Fast and energy-efficient neuromorphic deep learning with first-spike times
• Fractional-order spike-timing-dependent gradient descent for multi-layer spiking neural networks
• Efficient Deep Spiking Multi-Layer Perceptrons with Multiplication-Free Inference
• SpikE: spike-based embeddings for multi-relational graph data
• Energy efficiency analysis of Spiking Neural Networks for space applications
• Spike-based computation using classical recurrent neural networks
• Error-based or target-based? A unifying framework for learning in recurrent spiking networks
• Benchmarking Spiking Neural Network Learning Methods with Varying Locality
• Neuromorphic computing for attitude estimation onboard quadrotors
• Efficient Implementation of a Multi-Layer Gradient-Free Online-Trainable Spiking Neural Network on FPGA
• Supervised Training of Siamese Spiking Neural Networks with Earth Mover's Distance
• Dynamics and computation in mixed networks containing neurons that accelerate towards spiking
• Attacking the Spike: On the Transferability and Security of Spiking Neural Networks to Adversarial Examples
• Spiking mode-based neural networks
• Relational representation learning with spike trains
• A Realistic Simulation Framework for Analog/Digital Neuromorphic Architectures
• D-SELD: Dataset-Scalable Exemplar LCA-Decoder
• Adaptive Central Frequencies Locally Competitive Algorithm for Speech
• Neuronal architecture extracts statistical temporal patterns
• Three factor delay learning rules for spiking neural networks