An optimization scheme for segmented-memory neural network

被引:0
|
作者
Shan, Dongjing [1 ,2 ]
Zhang, Chao [2 ]
Nian, Yongjian [3 ]
机构
[1] Army Engn Univ, Lab Intelligent Informat Proc, Chongqing 400035, Peoples R China
[2] Peking Univ, Key Lab Machine Percept MOE, Beijing 100081, Peoples R China
[3] Third Mil Med Univ, Army Med Univ, Coll Biomed Engn & Imaging Med, Chongqing 400038, Peoples R China
关键词
Recurrent neural network; Human memory; Long-term memory; Temporal dependency; Temporal grouping; LONG-TERM DEPENDENCIES;
D O I
10.1016/j.neucom.2021.07.076
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Capture of long-term dependencies is a core task in sequence learning, and imitating the way of human memorization is a promising orientation. The existing algorithms can fractionate the sequence into segments with fixed length according to a prior knowledge, but the segmentation depends on the context and is difficult to assign length before network training. Thus in this paper, we propose a variant of segmented-memory neural network which can segment the sequence into arbitrary lengths and then perform cascade via a binarized mask of memory slots. For optimization of the network, we deduce an optimal mask theoretically, and then apply it in a novel scheme based on a sparsity regularizer. In experiments, we conduct ablation analysis and evaluation on some algorithmic or classification tasks, several models including the proposed one optimized by using lasso regularizer are adopted for comparison. Both of the fixed-and variable-length sequences are tested, and the results in different criteria have demonstrated the superiority of our proposed method. (c) 2021 Elsevier B.V. All rights reserved.
引用
收藏
页码:132 / 140
页数:9
相关论文
共 50 条
  • [1] Segmented-Memory Recurrent Neural Networks
    Chen, Jinmiao
    Chaudhari, Narendra S.
    IEEE TRANSACTIONS ON NEURAL NETWORKS, 2009, 20 (08): : 1267 - 1280
  • [2] Bidirectional segmented-memory recurrent neural network for protein secondary structure prediction
    Chen, J
    Chaudhari, NS
    SOFT COMPUTING, 2006, 10 (04) : 315 - 324
  • [3] Bidirectional segmented-memory recurrent neural network for protein secondary structure prediction
    J. Chen
    N.S. Chaudhari
    Soft Computing, 2006, 10 : 315 - 324
  • [4] Learning long-term dependencies in segmented-memory recurrent neural networks with backpropagation of error
    Gluege, Stefan
    Boeck, Ronald
    Palm, Guenther
    Wendemuth, Andreas
    NEUROCOMPUTING, 2014, 141 : 54 - 64
  • [5] SEGMENTED-MEMORY RECURRENT NEURAL NETWORKS VERSUS HIDDEN MARKOV MODELS IN EMOTION RECOGNITION FROM SPEECH
    Gluege, Stefan
    Boeck, Ronald
    Wendemuth, Andreas
    NCTA 2011: PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON NEURAL COMPUTATION THEORY AND APPLICATIONS, 2011, : 308 - 315
  • [6] A linear optimization neural network for associative memory
    Tao, Q
    Liu, X
    Cui, XX
    APPLIED MATHEMATICS AND COMPUTATION, 2005, 171 (02) : 1119 - 1128
  • [7] Memory Access Optimization of a Neural Network Accelerator Based on Memory Controller
    Wei, Rongshan
    Li, Chenjia
    Chen, Chuandong
    Sun, Guangyu
    He, Minghua
    ELECTRONICS, 2021, 10 (04) : 1 - 20
  • [8] Memory optimization at Edge for Distributed Convolution Neural Network
    Naveen, Soumyalatha
    Kounte, Manjunath R.
    TRANSACTIONS ON EMERGING TELECOMMUNICATIONS TECHNOLOGIES, 2022, 33 (12):
  • [9] Optimization of memory access for the convolutional neural network training
    Wang J.
    Hao Z.
    Li H.
    Xi'an Dianzi Keji Daxue Xuebao/Journal of Xidian University, 2020, 47 (02): : 98 - 107
  • [10] A SEGMENTED WEIGHTING SCHEME FOR STEEPEST ASCENT OPTIMIZATION
    WILLIAMS.WE
    FOWLER, WT
    AIAA JOURNAL, 1968, 6 (05) : 976 - &