[1]许邓艳、卢民荣.基于反向随机投影的神经网络改进算法[J].福建工程学院学报,2020,18(04):358-364.[doi:10.3969/j.issn.1672-4348.2020.04.010]
　XU Dengyan,LU Minrong.An improved neural network algorithm based on reverse random projection[J].Journal of FuJian University of Technology,2020,18(04):358-364.[doi:10.3969/j.issn.1672-4348.2020.04.010]
点击复制

基于反向随机投影的神经网络改进算法()
分享到：

《福建工程学院学报》[ISSN:2097-3853/CN:35-1351/Z]

卷:
第18卷
期数:
2020年04期
页码:
358-364
栏目:
出版日期:
2020-08-25

文章信息/Info

Title:
An improved neural network algorithm based on reverse random projection
作者:
许邓艳、卢民荣
福建工程学院应用技术学院
Author(s):
XU Dengyan;  LU Minrong
School of Applied Technology, Fujian University of Technology
关键词:
反向随机投影; 局部投影; 神经网络; 局部连接
Keywords:
reverse random projection;  local projection;  neural network;  local connection
分类号:
TP391.9
DOI:
10.3969/j.issn.1672-4348.2020.04.010
文献标志码:
A
摘要:
提出基于反向随机局部投影的神经网络效率改进算法，通过降低深度学习中的网络规模，重点解决了从“局部连接”到“全连接”和随机节点抽取时输入端节点信息丢失的问题，从而提升网络的效率。在算法中设置缩减参数，提升了算法的可伸展性，以适用于不同数据集的学习。通过数据集ISO-LET进行实验，结果表明，基于反向随机局部投影的神经网络效率改进算法的准确率、效率分别平均提升了3.48%和105.21%?在迭代20次的实验中进行了缩减参数调节实验，当参数设置为1.4时其准确率则优于传统全连接网络2.61%，效率提升了272.78%，具有明显的优势。
Abstract:
A neural network efficiency improvement algorithm based on reverse random local projection was proposed. By reducing the network size in deep learning, the problem of information loss of input nodes caused by “local connection” to “full connection” and random node extraction was solved, thus improving the efficiency of the network. The reduction parameters were set in the algorithm to improve the extensibility of the algorithm, so that it can be applied to the learning of different data sets. Then, experiments were conducted, using data set ISOLET. Results indicated that the accuracy and efficiency of the algorithm based on reverse random local projection were increased by 3.48% and 105.21% respectively. When echos were set as 20 the reduction parameter adjustment experiment was carried out, and when the parameter was set as 1.4, the accuracy was improved by 2.61% compared with the traditional fully-connected network, and the efficiency was improved by 272.78 %, which showed obvious advantages.

参考文献/References:

[1] 刘建伟, 王园方, 罗雄麟. 深度记忆网络研究进展[EB/OL]. 计算机学报, 2020:1-52[2020-08-11]. http:∥kns.cnki.net/kcms/detail/11. 1826.TP.20200114.1543.004.html. [2] HONAVAR V, UHR L. Artificial intelligence and neural networks-steps toward principled integration[J]. Science, 1994, 9(3): 545-548.[3] CHEN Y, EMER J, SZE V. Using dataflow to optimize energy efficiency of deep neural network accelerators[J]. IEEE Micro, 2017, 37(3): 12-21.[4] 朱虎明, 李佩, 焦李成, 等. 深度神经网络并行化研究综述[J]. 计算机学报, 2018, 41(8): 1861-1881.[5] DING S, SU C,YU J. An optimizing BP neural network algorithm based on genetic algorithm[J]. Artificial Intelligence Review, 2011, 36(2): 153-162.[6] 张禹, 李东升, 董小野, 等. 基于改进BP神经网络面向STEP-NC 2.5D制造特征的智能宏观工艺规划[J]. 机械工程学报, 2020, 56(1): 148-156. [7] ZHAO Y, ZHOU D, YAN H. An improved retrieval method of atmospheric parameter profiles based on the BP neural network[J]. Atmospheric Research, 2018, 213: 389-397.[8] 马力, 王永雄. 基于稀疏化双线性卷积神经网络的细粒度图像分类[J]. 模式识别与人工智能, 2019, 32(4): 336-344.[9] 韩敏, 穆云峰. 基于RBFLN网络的改进RBF神经网络学习算法[J]. 系统工程学报, 2008, 23(6): 764-768. [10] DENTON E, ZAREMBA W, BRUNA J, et al. Exploiting linear structure within convolutional networks for efficient evaluation[C]∥NIPS. MIT Press, 2014: 1269-1277.[11] LI X, PENG L, YAO X, et al. Long short-term memory neural network for air pollutant concentration predictions: method development and evaluation[J]. Environmental Pollution, 2017, 231: 997-1004.[12] ROY D, PANDA P, ROY K. Tree-CNN: a hierarchical deep convolutional neural network for incremental learning[J]. Neural Networks, 2020, 121: 148-160.[13] 王帅, 王维莹, 陈师哲, 等. 基于全局和局部信息的视频记忆度预测[J]. 软件学报, 2020, 31(7): 1969-1979.[14] 陈铬亮, 徐佳. 基于单词分类的归一化神经网络语言模型研究[J]. 福建工程学院学报, 2016, 14(4): 382-385.[15] RAHELI B, AALAMI M, EL-SHAFIE A, et al. Uncertainty assessment of the multilayer perceptron (MLP) neural network model with implementation of the novel hybrid MLP-FFA method for prediction of biochemical oxygen demand and dissolved oxygen: a case study of Langat River[J]. Environmental Earth Sciences, 2017, 76(14): 1-16. [16] ARJMANDZADEH Z, NAZEMI A, SAFI M. Solving multiobjective random interval programming problems by a capable neural network framework[J]. Applied Intelligence, 2019, 49(4): 1566-1579.[17] 冶忠林, 赵海兴, 张科, 等. 基于邻节点和关系模型优化的网络表示学习[J]. 计算机研究与发展, 2019, 56(12): 2562-2577. [18] 黄德根, 张云霞, 林红梅, 等. 基于规则推理网络的分类模型[J]. 软件学报, 2020, 31(4): 1063-1078.

更新日期/Last Update: 2020-08-25