Preprint
Abstract
In recent years, deep learning has been a revolution in the field of machine learning, for computer vision in particular. In this approach, a deep (multilayer) artificial neural network (ANN) is trained in a supervised manner using backpropagation. Huge amounts of labeled examples are required, but the resulting classification accuracy is truly impressive, sometimes outperforming humans. Neurons in an ANN are characterized by a single, static, continuous-valued activation. Yet biological neurons use discrete spikes to compute and transmit information, and the spike times, in addition to the spike rates, matter. Spiking neural networks (SNNs) are thus more biologically realistic than ANNs, and arguably the only viable option if one wants to understand how the brain computes. SNNs are also more hardware friendly and energy-efficient than ANNs, and are thus appealing for technology, especially for portable devices. However, training deep SNNs remains a challenge. Spiking neurons' transfer function is usually non-differentiable, which prevents using backpropagation. Here we review recent supervised and unsupervised methods to train deep SNNs, and compare them in terms of accuracy, but also computational cost and hardware friendliness. The emerging picture is that SNNs still lag behind ANNs in terms of accuracy, but the gap is decreasing, and can even vanish on some tasks, while the SNNs typically require much fewer operations.Keywords
All Related Versions
This publication has 115 references indexed in Scilit:
- Cellular and Circuit Mechanisms Shaping the Perceptual Properties of the Primate FoveaCell, 2017
- Spiking Deep Convolutional Neural Networks for Energy-Efficient Object RecognitionInternational Journal of Computer Vision, 2014
- Asynchronous spiking neural P systems with local synchronizationInformation Sciences, 2013
- Reservoir Computing TrendsKI - Künstliche Intelligenz, 2012
- Segmentation and Edge Detection Based on Spiking Neural Network ModelNeural Processing Letters, 2010
- Reservoir computing approaches to recurrent neural network trainingComputer Science Review, 2009
- Action Recognition Using a Bio-Inspired Feedforward Spiking NetworkInternational Journal of Computer Vision, 2009
- A network that uses few active neurones to code visual input predicts the diverse shapes of cortical receptive fieldsJournal of Computational Neuroscience, 2006
- Forming sparse representations by local anti-Hebbian learningBiological Cybernetics, 1990
- A decrease in firing threshold observed after induction of the EPSP-spike (E-S) component of long-term potentiation in rat hippocampal slicesExperimental Brain Research, 1990