基于FPGA的事件抽取模型与加速器的设计实现  

Design and implementation of event extraction model and accelerator based on FPGA

作　　者：韩哲 姜晶菲[1] 乔林波 窦勇[1] 许金伟 阚志刚 HAN Zhe;JIANG Jing-fei;QIAO Lin-bo;DOU Yong;XU Jin-wei;KAN Zhi-gang(School of Computer,National University of Defense Technology,Changsha 410073,China)

机构地区：[1]国防科技大学计算机学院,湖南长沙410073

出　　处：《计算机工程与科学》2020年第11期1941-1948,共8页Computer Engineering & Science

基　　金：国家重大专项计划(2018ZX01028101);预研项目(31513010602-1)。

摘　　要：事件抽取技术是实现特定信息快速提取的一种关键技术,可广泛应用于信息检索、情感分析等场景。中文事件抽取因需要考虑中文语言特性的问题,较英文事件抽取任务来说更为困难。基于当前前沿的英文事件抽取神经网络模型,提出了一种适合硬件计算的中文事件抽取神经网络模型CEE-DGCNN,其事件触发词分类在ACE2005中文语料库上实现了71.71%的F1值。并设计实现了相应的加速器,通过对数据的定点量化进一步优化了模型大小,其性能在Xilinx XCKU115 FPGA上达到了97 GOP/s,为CPU平台上性能的67倍。Event extraction technology is important to achieve the quickly extraction of specific information,and it can be widely used in information retrieval,sentiment analysis and other scenarios.Chinese event extraction is more difficult than English event extraction due to the characteristics of Chinese language.Based on the state-of-the-art English event extraction neural network model,a CEE-DGCNN(Chinese Event Extraction based on multi-layer Dilate Gated Convolutional Neural Network)is proposed,which is suitable for hardware implementation.CEE-DGCNN achieves 71.71%F1-score of trigger classification on the ACE2005 Chinese corpus.The accelerator of CEE-DGCNN is designed and implemented,and the model size is further optimized by quantization.The accelerator can achieve 97 GOP/s on the Xilinx XCKU115 FPGA,which is 67 times faster than CPU.

关 键 词：FPGA 事件抽取 膨胀门卷积神经网络 加速器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]