A Scatter-and-Gather Spiking Convolutional Neural Network on a Reconfigurable Neuromorphic Hardware.
A Scatter-and-Gather Spiking Convolutional Neural Network on a Reconfigurable Neuromorphic Hardware.
复制标题
DOI：
10.3389/fnins.2021.694170
复制
发表时间：
2021
影响因子：
4.3
通讯作者：
Huang R
中科院分区：
文献类型：
--
作者：
Zou C;Cui X;Kuang Y;Liu K;Wang Y;Wang X;Huang R
人工神经网络（ANN），如卷积神经网络（CNN），已经在许多机器学习任务中取得了最先进的成果。然而，使用大规模全精度CNN进行推理必然会导致大量的能量消耗和内存占用，这严重阻碍了它们在移动的和嵌入式系统上的部署。尖峰神经网络（SNN）受到生物大脑的高度启发，由于其在类脑学习方面的天然优势以及事件驱动通信和计算的高能效，正成为新的解决方案。然而，训练深度SNN仍然是一个主要挑战，并且ANN和SNN之间通常存在很大的准确性差距。在本文中，我们介绍了一种名为“分散和聚集”的硬件友好转换算法，将量化的ANN转换为无损的SNN，其中神经元与三元{− 1，0，1}突触权重连接。每个尖峰神经元都是无状态的，更像是原始的McCulloch和Pitts模型，因为它最多激发一个尖峰，并且需要在每个时间步重置。此外，我们开发了一个增量映射框架，以证明在可重构神经形态芯片上的高效网络部署。实验结果表明，我们的尖峰LeNet在MNIST和VGG-Net在CIFAR-10上分别获得99.37%和91.91%的分类准确率。此外，所提出的映射算法管理我们的神经形态芯片上的网络部署具有最大的资源效率和良好的灵活性。我们的片上四尖峰LeNet和VGG-Net在0.9 V、252 MHz下分别实现了0.38 ms/图像和3.24 ms/图像的实时推理速度，以及0.28 mJ/图像和2.3 mJ/图像的平均功耗，比传统GPU高出近两个数量级。
Artificial neural networks (ANNs), like convolutional neural networks (CNNs), have achieved the state-of-the-art results for many machine learning tasks. However, inference with large-scale full-precision CNNs must cause substantial energy consumption and memory occupation, which seriously hinders their deployment on mobile and embedded systems. Highly inspired from biological brain, spiking neural networks (SNNs) are emerging as new solutions because of natural superiority in brain-like learning and great energy efficiency with event-driven communication and computation. Nevertheless, training a deep SNN remains a main challenge and there is usually a big accuracy gap between ANNs and SNNs. In this paper, we introduce a hardware-friendly conversion algorithm called “scatter-and-gather” to convert quantized ANNs to lossless SNNs, where neurons are connected with ternary {−1,0,1} synaptic weights. Each spiking neuron is stateless and more like original McCulloch and Pitts model, because it fires at most one spike and need be reset at each time step. Furthermore, we develop an incremental mapping framework to demonstrate efficient network deployments on a reconfigurable neuromorphic chip. Experimental results show our spiking LeNet on MNIST and VGG-Net on CIFAR-10 datasetobtain 99.37% and 91.91% classification accuracy, respectively. Besides, the presented mapping algorithm manages network deployment on our neuromorphic chip with maximum resource efficiency and excellent flexibility. Our four-spike LeNet and VGG-Net on chip can achieve respective real-time inference speed of 0.38 ms/image, 3.24 ms/image, and an average power consumption of 0.28 mJ/image and 2.3 mJ/image at 0.9 V, 252 MHz, which is nearly two orders of magnitude more efficient than traditional GPUs.
登录
查看更多内容
DOI：
10.1109/82.842110
发表时间：
2000-05-01
影响因子：
4.4
作者：
Boahen, KA
通讯作者：
Boahen, KA
影响因子：
3.8
作者：
Abbott, LF
通讯作者：
Abbott, LF
影响因子：
8
作者：
Falez, Pierre;Tirilly, Pierre;Boulet, Pierre
通讯作者：
Boulet, Pierre
DOI：
10.1109/tcad.2015.2474396
发表时间：
2015-10-01
影响因子：
2.9
作者：
Akopyan, Filipp;Sawada, Jun;Modha, Dharmendra S.
通讯作者：
Modha, Dharmendra S.
影响因子：
6
作者：
Bohte, SM;Kok, JN;La Poutré, H
通讯作者：
La Poutré, H