SHF: Small: Collaborative Research: LDPD-Net: A Framework for Accelerated Architectures for Low-Density Permuted-Diagonal Deep Neural Networks

SHF：小型：协作研究：LDPD-Net：低密度置换对角深度神经网络加速架构框架

• 批准号: 1854737
• 负责人: Bo Yuan
• 金额: $ 22.5万
• 依托单位: Rutgers University New Brunswick
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-08-15 至 2021-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1854737&HistoricalAwards=false
• 关键词: SHF; Small; Collaborative; Research; LDPD

Deep learning has emerged as an important form of machine-learning where multiple layers of neural networks can learn the system function from available input-output data. Deep learning has outperformed traditional machine-learning algorithms based on feature engineering in fields such as image recognition, healthcare, and autonomous vehicles. These are widely used in cloud computing where large amount of computational resources are available. Deep neural networks are typically trained using graphic processing units (GPUs) or tensor processing units (TPUs). The training time and energy consumption grow with the complexity of the neural network. This project attempts to impose sparsity and regularity as constraints on the structure of the deep neural networks to reduce complexity and energy consumption by orders of magnitude, possibly at the expense of a slight degradation in the performance. The impacts lie in the formulation of a new family of structures for neural networks referred to as Low-Density Permuted Diagonal Network or LDPD-Net. The approach will enable the deployment of deep neural networks in energy-constrained and resource-constrained embedded platforms for inference tasks, including, but not limited to, unmanned vehicles/aerial systems, personalized healthcare, wearable and implantable devices, and mobile intelligent systems. In addition, the design methodology/techniques developed in this project can facilitate investigation of efficient computing of other matrix/tensor-based big data processing and analysis approaches. These approaches may also find applications in data-driven neuroscience and data-driven signal processing. In addition to graduate students, the project will involve undergraduates via senior design projects and research experiences for undergraduates. The results of the project will be disseminated to the broader community by publications, presentations, talks at various industries and other academic institutions. The main barriers to wide adoption of deep learning networks include computational resource constraints and energy consumption constraints. These barriers can be relaxed by imposing sparsity and regularity among different layers of the deep neural network. The proposed low-density permuted-diagonal (LDPD) network can lead to orders of magnitude reduction in computation complexity, storage space and energy consumption. The LDPD-Net will not be retrained by first training a regular network and then only retaining the weights corresponding to the LDPD-Net. Instead, the proposed network will be trained from scratch. The proposed LDPD-Net can enable scaling of the network for a specified computational platform. The proposed research has three thrusts: 1) develop novel resource-constrained and energy-constrained inference and training systems; 2) develop novel efficient hardware architectures that can fully exploit the advantages of the LDPD-Net to achieve high performance; and 3) perform novel software and hardware co-design and co-optimization to explore the design space of the LDPD-Net. Using these, the efficacy of the proposed LDPD-net will be validated and evaluated, via software implementations on high-performance systems, low-power embedded systems, and a hardware prototype on FPGA development boards.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

深度学习已成为机器学习的一种重要形式，多层神经网络可以从可用的输入输出数据中学习系统功能。在图像识别、医疗保健和自动驾驶汽车等领域，深度学习的性能优于基于特征工程的传统机器学习算法。它们广泛应用于有大量计算资源的云计算中。深度神经网络通常使用图形处理单元 (GPU) 或张量处理单元 (TPU) 进行训练。训练时间和能量消耗随着神经网络的复杂性而增长。该项目试图将稀疏性和规律性作为深度神经网络结构的约束，以将复杂性和能耗降低几个数量级，但可能会以性能略有下降为代价。其影响在于制定了一系列新的神经网络结构，称为低密度置换对角网络或 LDPD-Net。该方法将使深度神经网络能够在能量受限和资源受限的嵌入式平台中部署用于推理任务，包括但不限于无人驾驶车辆/航空系统、个性化医疗保健、可穿戴和植入设备以及移动智能系统。此外，该项目开发的设计方法/技术可以促进其他基于矩阵/张量的大数据处理和分析方法的高效计算的研究。这些方法也可能在数据驱动的神经科学和数据驱动的信号处理中找到应用。除了研究生之外，该项目还将通过本科生的高级设计项目和研究经验让本科生参与进来。该项目的成果将通过出版物、演讲、各行业和其他学术机构的演讲向更广泛的社区传播。广泛采用深度学习网络的主要障碍包括计算资源限制和能源消耗限制。通过在深度神经网络的不同层之间施加稀疏性和规律性，可以缓解这些障碍。所提出的低密度置换对角（LDPD）网络可以使计算复杂性、存储空间和能耗显着降低几个数量级。首先训练一个常规网络，然后只保留 LDPD-Net 对应的权重，不会对 LDPD-Net 进行重新训练。相反，所提出的网络将从头开始训练。所提出的 LDPD-Net 可以为指定的计算平台扩展网络。拟议的研究有三个重点：1）开发新颖的资源受限和能量受限的推理和训练系统； 2）开发新颖高效的硬件架构，可以充分利用LDPD-Net的优势来实现高性能； 3）进行新颖的软件和硬件协同设计和协同优化，以探索 LDPD-Net 的设计空间。使用这些，将通过高性能系统、低功耗嵌入式系统上的软件实现以及 FPGA 开发板上的硬件原型来验证和评估所提出的 LDPD-net 的功效。该奖项反映了 NSF 的法定使命，并已通过使用基金会的智力优点和更广泛的影响审查标准进行评估，认为值得支持。

项目成果

Global Placement Exploiting Soft 2D Regularity

利用软二维规律的全局布局

• 发表时间: 2022-01
• 期刊: International Symposium on Physical Design (ISPD
• 作者: Fang, D.;Zhang, B.;Hu, H.;Li, W.;Yuan, B.;Hu, J.
• 通讯作者: Hu, J.

Compressing Deep Neural Networks Using Toeplitz Matrix: Algorithm Design and Fpga Implementation

使用 Toeplitz 矩阵压缩深度神经网络：算法设计和 Fpga 实现

• DOI: 10.1109/icassp.2019.8683556
• 发表时间: 2019-05-12
• 期刊: ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
• 作者: Siyu Liao;Ashkan Samiee;Chunhua Deng;Y. Bai;Bo Yuan
• 通讯作者: Bo Yuan

Doubly Residual Neural Decoder: Towards Low-Complexity High-Performance Channel Decoding.

双残差神经解码器：迈向低复杂度高性能通道解码。

• 发表时间: 2021-01
• 期刊: Proceedings of the AAAI Conference on Artificial Intelligence
• 作者: Liao, S.;Deng, C.;Yin, M.;Yuan, B.
• 通讯作者: Yuan, B.

Hardware Acceleration of Persistent Homology Computation

持久同调计算的硬件加速

• DOI: 10.1007/978-3-030-33642-4_9
• 发表时间: 2019-10
• 期刊: International Workshop on Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention
• 作者: Wang, F;Deng, C;Yuan, B;Chen, C
• 通讯作者: Chen, C

Embedding Compression with Isotropic Iterative Quantization

具有各向同性迭代量化的嵌入压缩

• DOI: 10.1609/aaai.v34i05.6350
• 发表时间: 2020-01-11
• 期刊: ArXiv
• 作者: Siyu Liao;Jie Chen;Yanzhi Wang;Qinru Qiu;Bo Yuan
• 通讯作者: Bo Yuan