Most existing Spiking Neural Network (SNN) works state that SNNs may utilize temporal information dynamics of spikes. However, an explicit analysis of temporal information dynamics is still missing. In this paper, we ask several important questions for providing a fundamental understanding of SNNs: What are temporal information dynamics inside SNNs? How can we measure the temporal information dynamics? How do the temporal information dynamics affect the overall learning performance? To answer these questions, we estimate the Fisher Information of the weights to measure the distribution of temporal information during training in an empirical manner. Surprisingly, as training goes on, Fisher information starts to concentrate in the early timesteps. After training, we observe that information becomes highly concentrated in earlier few timesteps, a phenomenon we refer to as temporal information concentration. We observe that the temporal information concentration phenomenon is a common learning feature of SNNs by conducting extensive experiments on various configurations such as architecture, dataset, optimization strategy, time constant, and timesteps. Furthermore, to reveal how temporal information concentration affects the performance of SNNs, we design a loss function to change the trend of temporal information. We find that temporal information concentration is crucial to building a robust SNN but has little effect on classification accuracy. Finally, we propose an efficient iterative pruning method based on our observation on temporal information concentration. Code is available at https://github.com/Intelligent-Computing-Lab-Yale/Exploring-Temporal-Information-Dynamics-in-Spiking-Neural-Networks.
translated by 谷歌翻译
尖峰神经网络(SNNS)最近成为新一代的低功耗深神经网络,二进制尖峰在多个时间步中传达信息。 SNN的修剪非常重要,因为它们被部署在资源控制移动/边缘设备上。先前的SNN修剪作品的重点是浅SNN(2〜6层),但是,最深层的SNN(> 16层)是由最先进的SNN作品提出的,这很难与当前的修剪工作兼容。为了扩展针对深SNN的修剪技术,我们研究了彩票假说(LTH),该假说(LTH)指出,密集的网络包含较小的子网络(即获胜的票),这些子网与密集网络相当。我们对LTH的研究表明,获胜的门票始终存在于各种数据集和体系结构的深SNN中,可提供多达97%的稀疏性,而没有巨大的性能降级。但是,LTH的迭代搜索过程与SNN的多个时间段相结合时,带来了巨大的培训计算成本。为了减轻这种沉重的搜索成本,我们提出了早期(ET)票,从而从少量的时间步中找到重要的重量连接性。提出的ET票可以与常见的修剪技术无缝结合,以查找获胜门票,例如迭代级修剪(IMP)和早鸟(EB)门票。我们的实验结果表明,与IMP或EB方法相比,提出的ET票可将搜索时间缩短多达38%。
translated by 谷歌翻译
由于其异步,稀疏和二进制信息处理,尖峰神经网络(SNN)最近成为人工神经网络(ANN)的低功耗替代品。为了提高能源效率和吞吐量,可以在使用新兴的非挥发性(NVM)设备在模拟域中实现多重和蓄积(MAC)操作的回忆横梁上实现SNN。尽管SNN与回忆性横梁具有兼容性,但很少关注固有的横杆非理想性和随机性对SNN的性能的影响。在本文中,我们对SNN在非理想横杆上的鲁棒性进行了全面分析。我们检查通过学习算法训练的SNN,例如,替代梯度和ANN-SNN转换。我们的结果表明,跨多个时间阶段的重复横梁计算会导致错误积累,从而导致SNN推断期间的性能下降。我们进一步表明,经过较少时间步长培训的SNN在部署在磁带横梁上时可以更好地准确。
translated by 谷歌翻译
Spiking Neural Networks (SNNs) have gained huge attention as a potential energy-efficient alternative to conventional Artificial Neural Networks (ANNs) due to their inherent high-sparsity activation. Recently, SNNs with backpropagation through time (BPTT) have achieved a higher accuracy result on image recognition tasks than other SNN training algorithms. Despite the success from the algorithm perspective, prior works neglect the evaluation of the hardware energy overheads of BPTT due to the lack of a hardware evaluation platform for this SNN training algorithm. Moreover, although SNNs have long been seen as an energy-efficient counterpart of ANNs, a quantitative comparison between the training cost of SNNs and ANNs is missing. To address the aforementioned issues, in this work, we introduce SATA (Sparsity-Aware Training Accelerator), a BPTT-based training accelerator for SNNs. The proposed SATA provides a simple and re-configurable systolic-based accelerator architecture, which makes it easy to analyze the training energy for BPTT-based SNN training algorithms. By utilizing the sparsity, SATA increases its computation energy efficiency by $5.58 \times$ compared to the one without using sparsity. Based on SATA, we show quantitative analyses of the energy efficiency of SNN training and compare the training cost of SNNs and ANNs. The results show that, on Eyeriss-like systolic-based architecture, SNNs consume $1.27\times$ more total energy with sparsities when compared to ANNs. We find that such high training energy cost is from time-repetitive convolution operations and data movements during backpropagation. Moreover, to propel the future SNN training algorithm design, we provide several observations on energy efficiency for different SNN-specific training parameters and propose an energy estimation framework for SNN training. Code for our framework is made publicly available.
translated by 谷歌翻译
用尖峰神经网络(SNN)对基于事件的数据集开发神经形态智能最近引起了很多研究的关注。但是,基于事件的数据集的大小有限,使SNN易于过度拟合和不稳定的收敛性。以前的学术工作仍未探索这个问题。为了最大程度地减少这种泛化差距,我们提出了神经形态数据增强(NDA),这是一个专门针对基于事件的数据集设计的几何增强家族,目的是显着稳定SNN训练并减少训练和测试性能之间的概括差距。所提出的方法简单且与现有的SNN训练管道兼容。我们首次使用所提出的增强作用,证明了无监督的SNN对比度学习的可行性。我们对盛行的神经形态视觉基准进行了全面的实验,并表明NDA比以前的最新结果产生了实质性改进。例如,基于NDA的SNN分别在CIFAR10-DV和N-Caltech 101上获得了101%和13.7%的准确性增长。代码可在github https://github.com/intelligent-computing-lab-yale/nda_snn上找到。
translated by 谷歌翻译
尖峰神经网络(SNN)由于其固有的高表象激活而引起了传统人工神经网络(ANN)的势能有效替代品。但是,大多数先前的SNN方法都使用类似Ann的架构(例如VGG-NET或RESNET),这可以为SNN中二进制信息的时间序列处理提供亚最佳性能。为了解决这个问题,在本文中,我们介绍了一种新型的神经体系结构搜索(NAS)方法,以找到更好的SNN体系结构。受到最新的NAS方法的启发,这些方法从初始化时从激活模式中找到了最佳体系结构,我们选择了可以代表不同数据样本的不同尖峰激活模式的体系结构,而无需训练。此外,为了进一步利用尖峰之间的时间信息,我们在层之间搜索馈电的连接以及向后连接(即时间反馈连接)。有趣的是,我们的搜索算法发现的SNASNET通过向后连接实现了更高的性能,这表明设计SNN体系结构以适当使用时间信息的重要性。我们对三个图像识别基准进行了广泛的实验,我们表明SNASNET可以实现最新的性能,而时间段明显较低(5个时间段)。代码可在GitHub上找到。
translated by 谷歌翻译
二进制忆内横梁作为节能深度学习硬件加速器的巨大关注。尽管如此,由于横梁的模拟性质,它们遭受各种噪音。为了克服这些限制,最先前的作品训练重量参数,具有从横杆获得的噪声数据。然而,这些方法是无效的,因为难以在每个横杆具有大器件/电路电平变化的大容量制造环境中收集噪声数据。此外,我们认为即使这些方法有点提高了准确性,我们仍然存在改善余地。本文通过操纵输入二进制位编码而不是训练网络的重量,探讨了更广泛的方式以更广泛的方式缓解横杆噪声的新视角。我们首先在数学上示出,当表示相同量的信息时,随着二进制比特编码脉冲的数量增加,噪声减小。另外,我们提出了基于梯度的比特编码优化(GBO),其基于我们的深入分析,每层优化各层的不同数量的脉冲,即每个层具有不同的噪声灵敏度水平。所提出的异构层面编码方案具有低计算成本的高噪声鲁棒性。我们对公共基准数据集的实验结果表明,GBO在严重的噪声场景中提高了〜5-40%的分类精度。
translated by 谷歌翻译
我们如何为神经系统带来隐私和能效?在本文中,我们提出了PrivateNN,旨在从预先训练的ANN模型构建低功耗尖峰神经网络(SNNS),而不会泄漏包含在数据集中的敏感信息。在这里,我们解决两种类型的泄漏问题:1)当网络在Ann-SNN转换过程中访问真实训练数据时,会导致数据泄漏。 2)当类相关的特征可以从网络参数重建时,会导致类泄漏。为了解决数据泄漏问题,我们从预先培训的ANN生成合成图像,并使用所生成的图像将ANN转换为SNNS。然而,转换的SNNS仍然容易受到类泄漏的影响,因为权重参数相对于ANN参数具有相同的(或缩放)值。因此,通过训练SNNS,通过训练基于时间尖峰的学习规则来加密SNN权重。使用时间数据更新权重参数使得SNN难以在空间域中解释。我们观察到,加密的私人没有消除数据和类泄漏问题,略微的性能下降(小于〜2),与标准ANN相比,与标准ANN相比的显着的能效增益(约55倍)。我们对各种数据集进行广泛的实验,包括CiFar10,CiFar100和Tinyimagenet,突出了隐私保留的SNN培训的重要性。
translated by 谷歌翻译
由于稀疏,异步和二进制事件(或尖峰)驱动加工,尖峰神经网络(SNNS)最近成为深度学习的替代方案,可以在神经形状硬件上产生巨大的能效益。然而,从划痕训练高精度和低潜伏期的SNN,患有尖刺神经元的非微弱性质。要在SNNS中解决此培训问题,我们重新批准批量标准化,并通过时间(BNTT)技术提出时间批量标准化。大多数先前的SNN工程到现在忽略了批量标准化,认为它无效地训练时间SNN。与以前的作品不同,我们提出的BNTT沿着时轴沿着时间轴解耦的参数,以捕获尖峰的时间动态。在BNTT中的时间上不断发展的可学习参数允许神经元通过不同的时间步长来控制其尖峰率,从头开始实现低延迟和低能量训练。我们对CiFar-10,CiFar-100,微小想象特和事件驱动的DVS-CIFAR10数据集进行实验。 BNTT允许我们首次在三个复杂的数据集中培训深度SNN架构,只需25-30步即可。我们还使用BNTT中的参数分布提前退出算法,以降低推断的延迟,进一步提高了能量效率。
translated by 谷歌翻译
Machine Translation (MT) system generally aims at automatic representation of source language into target language retaining the originality of context using various Natural Language Processing (NLP) techniques. Among various NLP methods, Statistical Machine Translation(SMT). SMT uses probabilistic and statistical techniques to analyze information and conversion. This paper canvasses about the development of bilingual SMT models for translating English to fifteen low-resource Indian Languages (ILs) and vice versa. At the outset, all 15 languages are briefed with a short description related to our experimental need. Further, a detailed analysis of Samanantar and OPUS dataset for model building, along with standard benchmark dataset (Flores-200) for fine-tuning and testing, is done as a part of our experiment. Different preprocessing approaches are proposed in this paper to handle the noise of the dataset. To create the system, MOSES open-source SMT toolkit is explored. Distance reordering is utilized with the aim to understand the rules of grammar and context-dependent adjustments through a phrase reordering categorization framework. In our experiment, the quality of the translation is evaluated using standard metrics such as BLEU, METEOR, and RIBES
translated by 谷歌翻译