基于权重活性评价的循环神经网络模型  

RECURRENT NEURAL NETWORK MODEL BASED ON WEIGHT ACTIVITY EVALUATION

作　　者：张承 郑明 胡雨阳 夏定纯 Zhang Cheng;Zheng Ming;Hu Yuyang;Xia Dingchun(School of Mathematics and Computer Science,Wuhan Textile University,Wuhan 430200,Hubei,China)

机构地区：[1]武汉纺织大学数学与计算机学院,湖北武汉430200

出　　处：《计算机应用与软件》2024年第6期296-304,319,共10页Computer Applications and Software

摘　　要：针对LSTM等循环神经网络存在着结构复杂和参数冗余等问题,对循环神经网络的结构做了相关的研究和分析,提出一种权重活性评价算法,对网络的基本单元进行活性评价和结构的筛选,以提高循环神经网络的结构合理性,降低网络参数的计算量。通过对心律失常数据的实验与测试,分析LSTM网络权重活性的差异以及权重与梯度的变化特征,实验结果表明,运用该算法能够较好地优化循环神经网络的结构,并降低网络参数的冗余。Aimed at the complex structure and parameter redundancy of recurrent neural networks such as LSTM,the structure of recurrent neural networks is studied and analyzed.In order to improve the structural rationality of the recurrent neural network and reduce the amount of calculation of network parameters,a weight activity evaluation algorithm is proposed that evaluates the activity of the basic unit of the network and screens its structure.Through experiments and tests on arrhythmia data,the differences in the weight activity of the LSTM network and the change characteristics of weights and gradients were analyzed.The experimental results show that this algorithm can better optimize the structure of the recurrent neural network and reduce the redundancy of network parameters.

关 键 词：权重 活性 评价 参数冗余 循环神经网络 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]