A brain-inspired spiking neural network model with temporal encoding and learning

Yu, Qiang; Tang, Huajin; Tan, Kay Chen; Yu, Haoyong

doi:10.1016/j.neucom.2013.06.052

Cited by 114 publications

(53 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The first theory is based on the idea of "temporal code" [8,10,24,25] and goes into the spike -trains structure while the second referred to as "rate code" theory [9,15,19,23,24] assumes that the neural code is embedded in the spike frequency, defined as the number of spikes emitted per second. The temporal coding mechanism, which builds a relationship of temporal process between the output firing patterns and the inputs of the nervous system, has received much attention [7,11,13].…”

Section: Introductionmentioning

confidence: 99%

Temporal code versus rate code for binary Information Sources

2016

View full text Add to dashboard Cite

Neuroscientists formulate very different hypotheses about the nature of neural code. At one extreme, it has been argued that neurons encode information in relatively slow changes of individual spikes arriving rates (rates codes) and the irregularity in the spike trains reflects noise in the system, while in the other ex-* Corresponding author Email address: jszczepa@ippt.pan.pl (Janusz Szczepanski ) and rate codes. On the contrary, it turns out that for large enough jumping parameter this quotient is non-monotonic function of firing rate and it exhibits a global maximum, i.e. in this case there exists the optimal firing rate. Moreover, there is no one-to-one relation between information and firing rates, so the temporal and rate codes differ qualitatively. This leads to the observation that the behavior of the quotient of information and firing rates for large jumping parameter is especially important in the context of bursting phenomena.

show abstract

Section: Introductionmentioning

confidence: 99%

Temporal code versus rate code for binary Information Sources

2016

View full text Add to dashboard Cite

show abstract

“…The Time Coding method encodes information into spike emission date, which allows to use only one spike per input pixel (for image processing example).Unlike rate coding, a gray-scale image is encoded by signals holding only one spike per pixel. This latter is emitted in a time that is inversely proportional to the pixel's intensity [39] [44], as depicted in figure 1b. In this model, spikes are dependent on each other, because their arrival times can be interpreted only relatively to other spikes.…”

Section: Time Codingmentioning

confidence: 99%

Design Space Exploration of Hardware Spiking Neurons for Embedded Artificial Intelligence

2020

View full text Add to dashboard Cite

A B S T R A C TMachine learning is yielding unprecedented interest in research and industry, due to recent success in many applied contexts such as image classification and object recognition. However, the deployment of these systems requires huge computing capabilities, thus making them unsuitable for embedded systems. To deal with this limitation, many researchers are investigating brain-inspired computing, which would be a perfect alternative to the conventional Von Neumann architecture based computers (CPU/GPU) that meet the requirements for computing performance, but not for energy-efficiency. Therefore, neuromorphic hardware circuits that are adaptable for both parallel and distributed computations need to be designed. In this paper, we focus on Spiking Neural Networks (SNNs) with a comprehensive study of information coding methods and hardware exploration. In this context, we propose a framework for neuromorphic hardware design space exploration, which allows to define a suitable architecture based on application-specific constraints and starting from a wide variety of possible architectural choices. For this framework, we have developed a behavioral level simulator for neuromorphic hardware architectural exploration named NAXT. Moreover, we propose modified versions of the standard Rate Coding technique to make trade-offs with the Time Coding paradigm, which is characterized by the low number of spikes propagating in the network. Thus, we are able to reduce the number of spikes while keeping the same neuron's model, which results in an SNN with fewer events to process. By doing so, we seek to reduce the amount of power consumed by the hardware. Furthermore, we present three neuromorphic hardware architectures in order to quantitatively study the implementation of SNNs. One of these architectures integrates a novel hybrid structure: a highly-parallel computation core for most solicited layers, and time-multiplexed computation units for deeper layers. These architectures are derived from a novel funnel-like Design Space Exploration framework for neuromorphic hardware.

show abstract

“…Thereafter, the signal is subjected to a Level 4 Daubechies wavelet decomposition of 4th order. The wavelet decomposition yields signals in five sub-bands, namely Delta (0-4 Hz), Theta (4-7 Hz), Alpha (8-15 Hz), Beta (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) and Gamma . Using the band-limited EEG signal and the sub-bands, the extracted features are: standard deviation computed from the band-limited EEG, alpha, beta and gamma subbands; the correlation dimension computed from the alpha, beta and gamma sub-bands; the largest Lyapunov exponent computed from the band-limited EEG and alpha sub-band.…”

Section: Performance Of Sresn On Epilepsy Detection Problemmentioning

confidence: 99%

“…Tempotron [15,16] works by minimizing the difference between the maximum potential reached by the neuron and the threshold potential. These works employ a two layered architecture with leaky integrate-and-fire spiking neurons.…”

Section: Introductionmentioning

confidence: 99%