T2FSNN: Deep Spiking Neural Networks with Time-to-first-spike Coding

Park, Seong Sik; Kim, Seijoon; Na, Byunggook; Yoon, Sungroh

doi:10.1109/dac18072.2020.9218689

Cited by 88 publications

(49 citation statements)

References 43 publications

(75 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the previous work, different neural coding schemes were compared in terms of classification accuracy, latency, the number of spikes, and energy during inference ( Park et al, 2020 ). The comparison revealed that TTFS coding won against the other coding schemes in classification and computational performance.…”

Section: Comparison and Discussionmentioning

confidence: 99%

“…Time to first spike coding was discovered to encode information for fast responses within a few milliseconds, like tactile stimulus ( Johansson and Birznieks, 2004 ), by using the first spikes. Park et al (2020) proposed a fast and energy-efficient TTFS coding scheme that used an exponential-decaying dynamic threshold to convert input pixels to the first-spike patterns. The larger an input pixel is, the more information it carries, and the earlier it emits a spike.…”

Section: Background and Methodsmentioning

confidence: 99%

“…Many experiments have pointed out the significance of the first spikes in various parts of our nervous system, such as retina, auditory systems, and tactile afferents ( Ponulak and Kasinski, 2011 ). Various works have reported that applying TTFS coding scheme in SNNs can significantly reduce the number of spikes and improve inference speed ( Rueckauer and Liu, 2018 ; Oh et al, 2020 ; Park et al, 2020 ). Phase coding encodes information in spike patterns whose phases are correlated with internally generated background oscillation rhythms, which has been experimentally observed in the hippocampus and olfactory system ( O’Keefe and Recce, 1993 ; Laurent, 1996 ).…”

Section: Introductionmentioning

confidence: 99%

“…Recent studies compared different neural coding schemes in inference performance in SNNs converted from well-trained deep neural networks (DNNs) ( Park et al, 2019 , 2020 ). The TTFS-based temporal coding was demonstrated to exhibit the highest classification accuracy, fastest inference speed, and lowest energy consumption.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Neural Coding in Spiking Neural Networks: A Comparative Study for Robust Neuromorphic Systems

et al. 2021

View full text Add to dashboard Cite

Various hypotheses of information representation in brain, referred to as neural codes, have been proposed to explain the information transmission between neurons. Neural coding plays an essential role in enabling the brain-inspired spiking neural networks (SNNs) to perform different tasks. To search for the best coding scheme, we performed an extensive comparative study on the impact and performance of four important neural coding schemes, namely, rate coding, time-to-first spike (TTFS) coding, phase coding, and burst coding. The comparative study was carried out using a biological 2-layer SNN trained with an unsupervised spike-timing-dependent plasticity (STDP) algorithm. Various aspects of network performance were considered, including classification accuracy, processing latency, synaptic operations (SOPs), hardware implementation, network compression efficacy, input and synaptic noise resilience, and synaptic fault tolerance. The classification tasks on Modified National Institute of Standards and Technology (MNIST) and Fashion-MNIST datasets were applied in our study. For hardware implementation, area and power consumption were estimated for these coding schemes, and the network compression efficacy was analyzed using pruning and quantization techniques. Different types of input noise and noise variations in the datasets were considered and applied. Furthermore, the robustness of each coding scheme to the non-ideality-induced synaptic noise and fault in analog neuromorphic systems was studied and compared. Our results show that TTFS coding is the best choice in achieving the highest computational performance with very low hardware implementation overhead. TTFS coding requires 4x/7.5x lower processing latency and 3.5x/6.5x fewer SOPs than rate coding during the training/inference process. Phase coding is the most resilient scheme to input noise. Burst coding offers the highest network compression efficacy and the best overall robustness to hardware non-idealities for both training and inference processes. The study presented in this paper reveals the design space created by the choice of each coding scheme, allowing designers to frame each scheme in terms of its strength and weakness given a designs’ constraints and considerations in neuromorphic systems.

show abstract

Section: Comparison and Discussionmentioning

confidence: 99%

Section: Background and Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Neural Coding in Spiking Neural Networks: A Comparative Study for Robust Neuromorphic Systems

et al. 2021

View full text Add to dashboard Cite

show abstract

“…The most complex part of the above-mentioned SNN is implementing the exponentially decaying synaptic current model in hardware. In several previous works, TTFS data was trained on the network using the unconventional current model, but most models [21] were too complex to implement in hardware. In this paper, we propose a method for reading FET- When a triangle pulse that decreases linearly in the timedomain is applied to the gate of the device, the subthreshold current can effectively represent an exponentially decaying model.…”

Section: Ers @mentioning

confidence: 99%

Spiking Neural Networks With Time-to-First-Spike Coding Using TFT-Type Synaptic Device Model

Soo-Chang

Woo

et al. 2021

IEEE Access

View full text Add to dashboard Cite

In hardware-based spiking neural networks (SNNs), the conversion of analog input data into the arrival time of an input pulse is regarded as a good candidate for the encoding method due to its bioplausibility and power-efficiency. In this work, we trained an SNN encoded by time to first spike (TTFS) and performed an inference process using the behavior of the fabricated TFT-type flash synaptic device. The exponentially decaying synaptic current model required in the inference process was implemented by reading devices in the subthreshold region using triangle pulses. In a high-level system simulation, the TTFS-SNN (two-layer MLP with 512 hidden neurons) reached a high accuracy of 97.94%. Compared to conventional rate-encoded SNNs, TTFS-SNN made 2.9 times faster judgment and consumed ~10 times less energy in the inference process. Additionally, to use the network in a more stable condition, we propose a method to operate it using a rectangle pulse in the saturation region of the synaptic device. The distortion caused by this approximation was minimized by shortening the pulse width. As a result, the modified inference system showed an accuracy of 97.36%, and the prediction time and energy consumption were reduced 3.97-and 83.04-times when compared to those of the rate-SNN. Finally, we analyzed the sensitivity of the network performance due to unexpected issues that may occur in the hardware system and thus explained the competitiveness of the proposed synaptic behavior in the saturation region.

show abstract

Real Spike: Learning Real-Valued Spikes for Spiking Neural Networks

Guo

Zhang

Chen

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Brain-inspired spiking neural networks (SNNs) have recently drawn more and more attention due to their event-driven and energyefficient characteristics. The integration of storage and computation para-digm on neuromorphic hardwares makes SNNs much different from Deep Neural Networks (DNNs). In this paper, we argue that SNNs may not benefit from the weight-sharing mechanism, which can effectively reduce parameters and improve inference efficiency in DNNs, in some hardwares, and assume that an SNN with unshared convolution kernels could perform better. Motivated by this assumption, a training-inference decoupling method for SNNs named as Real Spike is proposed, which not only enjoys both unshared convolution kernels and binary spikes in inference-time but also maintains both shared convolution kernels and Real-valued Spikes during training. This decoupling mechanism of SNN is realized by a re-parameterization technique. Furthermore, based on the training-inference-decoupled idea, a series of different forms for implementing Real Spike on different levels are presented, which also enjoy shared convolutions in the inference and are friendly to both neuromorphic and non-neuromorphic hardware platforms. A theoretical proof is given to clarify that the Real Spike-based SNN network is superior to its vanilla counterpart. Experimental results show that all different Real Spike versions can consistently improve the SNN performance. Moreover, the proposed method outperforms the state-of-the-art models on both non-spiking static and neuromorphic datasets.

show abstract

T2FSNN: Deep Spiking Neural Networks with Time-to-first-spike Coding

Cited by 88 publications

References 43 publications

Neural Coding in Spiking Neural Networks: A Comparative Study for Robust Neuromorphic Systems

Neural Coding in Spiking Neural Networks: A Comparative Study for Robust Neuromorphic Systems

Spiking Neural Networks With Time-to-First-Spike Coding Using TFT-Type Synaptic Device Model

Real Spike: Learning Real-Valued Spikes for Spiking Neural Networks

Contact Info

Product

Resources

About