Survey on Accelerating Neural Network with Hardware
-
摘要: 人工神经网络目前广泛应用于人工智能的应用当中，如语音助手、图像识别和自然语言处理等.随着神经网络愈加复杂，计算量也急剧上升，传统的通用芯片在处理复杂神经网络时受到了带宽和能耗的限制，人们开始改进通用芯片的结构以支持神经网络的有效处理.此外，研发专用加速芯片也成为另一条加速神经网络处理的途径.与通用芯片相比，它能耗更低，性能更高.通过介绍目前通用芯片和专用芯片对神经网络所作的支持，了解最新神经网络硬件加速平台设计的创新点和突破口.具体来说，主要概述了神经网络的发展，讨论各类通用芯片为支持神经网络所作的改进，其中包括支持低精度运算和增加一个加速神经网络处理的计算模块.然后从运算结构和存储结构的角度出发，归纳专用芯片在体系结构上所作的定制设计，另外根据神经网络中各类数据的重用总结了各个神经网络加速器所采用的数据流.最后通过对已有加速芯片的优缺点分析，给出了神经网络加速器未来的设计趋势和挑战.Abstract: Artificial neural networks are widely used in artificial intelligence applications such as voice assistant, image recognition and natural language processing. With the rise of complexity of the application, the computational complexity has also increased dramatically. The traditional general-purpose processor is limited by the memory bandwidth and energy consumption when dealing with the complex neural network. People began to improve the architecture of the general-purpose processors to support the efficient processing of the neural network. In addition, the development of special-purpose accelerators becomes another way to accelerate processing of neural network. Compared with the general-purpose processor, it has lower energy consumption and higher performance. The article aims to introduce the designs from current general-purpose processors and special-purpose accelerators for supporting the neural network. It also summarizes the latest design innovation and breakthrough of the neural network acceleration platforms. In particular, the article provides an overview of the neural network and discusses the improvements made by various general-purpose chips to support neural networks, which include supporting low-precision operations and adding a calculation module to speed up neural network processing. Then from the viewpoint of the computational structure and storage structure, the article summarizes the customized designs of special-purpose accelerators, and describes the dataflow used by the neural network chips based on the reuse of various types of the data in the neural network. Through analyzing the advantages and disadvantages of these solutions, the article puts forward the future design trend and challenge of the neural network accelerator.
-
-
期刊类型引用(10)
1. 杨秀璋，彭国军，刘思德，田杨，李晨光，傅建明. 面向APT攻击的溯源和推理研究综述. 软件学报. 2025(01): 203-252 . 
百度学术
2. 申国霞，常鑫. 基于可信密码模块的网络信道潜在攻击挖掘. 信息技术. 2023(10): 152-156+162 . 百度学术
3. 谢峥，路广平，付安民. 一种可扩展的实时多步攻击场景重构方法. 信息安全研究. 2023(12): 1173-1179 . 百度学术
4. 黄维贵，孙怡峰，欧旺，王玉宾. 基于不确定攻击图的违规外联风险分析. 信息工程大学学报. 2022(05): 570-577 . 百度学术
5. 王文娟，杜学绘，单棣斌. 基于动态概率攻击图的云环境攻击场景构建方法. 通信学报. 2021(01): 1-17 . 百度学术
6. 潘亚峰，朱俊虎，周天阳. APT攻击场景重构方法综述. 信息工程大学学报. 2021(01): 55-60+80 . 百度学术
7. 罗智勇，杨旭，刘嘉辉，许瑞. 基于贝叶斯攻击图的网络入侵意图分析模型. 通信学报. 2020(09): 160-169 . 百度学术
8. 王硕，王建华，汤光明，裴庆祺，张玉臣，刘小虎. 一种智能高效的最优渗透路径生成方法. 计算机研究与发展. 2019(05): 929-941 . 本站查看
9. 吴东，郭春，申国伟. 一种基于多因素的告警关联方法. 计算机与现代化. 2019(06): 30-37 . 百度学术
10. 韩宜轩，秦元庆. 基于因果关联的电力工控系统攻击场景还原. 信息技术. 2019(08): 41-44+48 . 百度学术
其他类型引用(12)
计量
- 文章访问数: 2952
- HTML全文浏览量: 11
- PDF下载量: 2224
- 被引次数: 22
下载:
