基于无标度网络的类脑储备池拓扑设计

doi:10.3785/j.issn.1008-973X.2025.07.006

浙江大学学报(工学版)

2025, Vol. 59

Issue (7): 1385-1393 DOI: 10.3785/j.issn.1008-973X.2025.07.006

计算机技术与控制工程

基于无标度网络的类脑储备池拓扑设计

刘瑄昀1(

),闫莹1,於志勇1,2,3,黄昉菀1,2,3,*(

)

1. 福州大学计算机与大数据学院，福建福州 350108
2. 福州大学福建省网络计算与智能信息处理重点实验室，福建福州 350108
3. 大数据智能教育部工程研究中心，福建福州 350108

Topological design of brain-like reservoir based on scale-free network

Xuanyun LIU1(

),Ying YAN1,Zhiyong YU1,2,3,Fangwan HUANG1,2,3,*(

)

1. College of Computer and Data Sciences, Fuzhou University, Fuzhou 350108, China
2. Fujian Key Laboratory of Network Computing and Intelligent Information Processing, Fuzhou University, Fuzhou 350108, China
3. Engineering Research Center of Big Data Intelligence, Ministry of Education, Fuzhou 350108, China

全文: PDF(1271 KB) HTML

摘要：

为了优化回声状态网络（ESN）储备池的设计，应用基于随机矩阵理论的Chung-Lu（CL）构造算法生成灵活且高效的无标度网络. 针对在构建过程中出现的度值偏差，使用随机剪枝或度值剪枝改进无标度网络. 通过模拟随机攻击或针对性攻击来提高储备池的鲁棒性. 实验结果表明，加入剪枝机制的CL算法构造了具有幂律性质的无标度网络，构建速度和预测性能均明显优于基线算法，随机剪枝的效果优于度值剪枝. 相比基线算法的最优结果，基于随机剪枝的CL算法构建的ESN的运行时间和预测误差最少降低了14.2%和10.6%.

关键词： 无标度网络; 类脑储备池; 剪枝机制; 回声状态网络; 时间序列预测

Abstract:

To optimize the reservoir design of echo state networks (ESN), the Chung-Lu (CL) construction algorithm based on random matrix theory was applied to generate flexible and efficient scale-free networks. In response to the degree deviation that occurs during the construction process, random pruning or degree pruning was used to improve the scale-free network. The robustness of the reservoir was improved by simulating random or targeted attacks. Experimental results show that the CL algorithm with a pruning mechanism not only constructs a scale-free network with power-law property, but also has significantly better construction speed and prediction performance than the baseline algorithms, and the effect of random pruning was better than that of degree pruning. Compared with the optimal result of the baseline algorithm, the running time and prediction error of ESN constructed by the CL algorithm based on random pruning were at least reduced by 14.2% and 10.6%.

Key words: scale-free network brain-like reservoir pruning mechanism echo state network time series prediction

收稿日期: 2024-09-03 出版日期: 2025-07-25

CLC:

TP 393

基金资助: 国家自然科学基金资助项目（62332014）；福建省引导性资助项目（2020H0008）；福建省促进海洋与渔业产业高质量发展专项资金（FJHYF-ZH-2023-02）.

通讯作者: 黄昉菀 E-mail: 3167131927@qq.com;hfw@fzu.edu.cn

作者简介: 刘瑄昀（2001—），女，硕士生，从事机器学习研究. orcid.org/0009-0009-0489-1502. E-mail：3167131927@qq.com

	服务
	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	作者相关文章
	刘瑄昀
	闫莹
	於志勇
	黄昉菀

引用本文:

刘瑄昀,闫莹,於志勇,黄昉菀. 基于无标度网络的类脑储备池拓扑设计[J]. 浙江大学学报(工学版), 2025, 59(7): 1385-1393.

Xuanyun LIU,Ying YAN,Zhiyong YU,Fangwan HUANG. Topological design of brain-like reservoir based on scale-free network. Journal of ZheJiang University (Engineering Science), 2025, 59(7): 1385-1393.

链接本文:

https://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2025.07.006 或 https://www.zjujournals.com/eng/CN/Y2025/V59/I7/1385

图 1 改进Chung-Lu算法生成无标度网络的节点度分布

图 2 MSO数据集的前10 000个数据点（步长为0.1）

图 3 不同幂律指数下经不同机制剪枝后的节点度分布

图 4 不同机制下理想平均度值与真实平均度值的对比

图 5 不同机制下不同幂律指数的富人俱乐部系数的对比

表 1 基于Chung-Lu算法的储备池稀疏度

表 2 基于Chung-Lu算法的回声状态网络均方误差

表 3 基于Chung-Lu算法的回声状态网络平均绝对百分比误差

表 4 基于Chung-Lu算法的回声状态网络运行时间

表 5 不同数据集下不同方法构建的回声状态网络预测性能对比

1	FOLGHERAITER M, YSKAK A, YESSIRKEPOV S One-shot bipedal robot dynamics identification with a reservoir-based RNN[J]. IEEE Access, 2023, 11: 50180- 50194 doi: 10.1109/ACCESS.2023.3277977
2	WANG H, LONG X, LIU X X FastESN: fast echo state network[J]. IEEE Transactions on Neural Networks and Learning Systems, 2023, 34 (12): 10487- 10501 doi: 10.1109/TNNLS.2022.3167466
3	LIN J, CHUNG F L, WANG S A fast parametric and structural transfer leaky integrator echo state network for reservoir computing[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2024, 54 (5): 3257- 3269 doi: 10.1109/TSMC.2024.3358567
4	AKIYAMA T, TANAKA G Computational efficiency of multi-step learning echo state networks for nonlinear time series prediction[J]. IEEE Access, 2022, 10: 28535- 28544 doi: 10.1109/ACCESS.2022.3158755
5	VIEHWEG J, WORTHMANN K, MÄDER P Parameterizing echo state networks for multi-step time series prediction[J]. Neurocomputing, 2023, 522: 214- 228 doi: 10.1016/j.neucom.2022.11.044
6	HUANG L L, ZHANG Y, XIANG J H, et al Extreme multistability in a hopfield neural network based on two biological neuronal systems[J]. IEEE Transactions on Circuits and Systems II: Express Briefs, 2022, 69 (11): 4568- 4572
7	BEHERA R K, NAIK D, RAMESH D, et al MR-IBC: mapreduce-based incremental betweenness centrality in large-scale complex networks[J]. Social Network Analysis and Mining, 2020, 10 (1): 25 doi: 10.1007/s13278-020-00636-9
8	ALESSANDRETTI L, NATERA OROZCO L G, SABERI M, et al Multimodal urban mobility and multilayer transport networks[J]. Environment and Planning B: Urban Analytics and City Science, 2023, 50 (8): 2038- 2070 doi: 10.1177/23998083221108190
9	杨旭华, 王磊, 叶蕾, 等基于节点相似性和网络嵌入的复杂网络社区发现算法[J]. 计算机科学, 2022, 49 (3): 121- 128 YANG Xuhua, WANG Lei, YE Lei, et al Complex network community detection algorithm based on node similarity and network embedding[J]. Computer Science, 2022, 49 (3): 121- 128 doi: 10.11896/jsjkx.210200009
10	楼洋, 李均利, 李升, 等复杂网络能控性鲁棒性研究进展[J]. 自动化学报, 2022, 48 (10): 2374- 2391 LOU Yang, LI Junli, LI Sheng, et al Recent progress in controllability robustness of complex networks[J]. Acta Automatica Sinica, 2022, 48 (10): 2374- 2391
11	KAWAI Y, PARK J, ASADA M A small-world topology enhances the echo state property and signal propagation in reservoir computing[J]. Neural Networks, 2019, 112: 15- 23 doi: 10.1016/j.neunet.2019.01.002
12	WANG Q, DU C, GENG Y, et al Historical payoff can not overcome the vaccination dilemma on Barabási–Albert scale-free networks[J]. Chaos, Solitons and Fractals, 2020, 130: 109453 doi: 10.1016/j.chaos.2019.109453
13	CHENG J, WU X, ZHOU M, et al A novel method for detecting new overlapping community in complex evolving networks[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2019, 49 (9): 1832- 1844 doi: 10.1109/TSMC.2017.2779138
14	TSIOTAS D Detecting differences in the topology of scale-free networks grown under time-dynamic topological fitness[J]. Scientific Reports, 2020, 10: 10630 doi: 10.1038/s41598-020-67156-6
15	SIDOROV S, MIRONOV S, GRIGORIEV A, et al An investigation into the trend stationarity of local characteristics in media and social networks[J]. Systems, 2022, 10 (6): 249 doi: 10.3390/systems10060249
16	WANG B, ZHAO Y, HU F, et al Anomaly detection with subgraph search and vertex classification preprocessing in Chung-Lu random networks[J]. IEEE Transactions on Signal Processing, 2018, 66 (20): 5255- 5268 doi: 10.1109/TSP.2018.2866847
17	GLOS A Spectral similarity for Barabási–Albert and Chung-Lu models[J]. Physica A: Statistical Mechanics and Its Applications, 2019, 516: 571- 578 doi: 10.1016/j.physa.2018.10.050
18	FASINO D, TONETTO A, TUDISCO F Generating large scale-free networks with the Chung-Lu random graph model[J]. Networks, 2021, 78 (2): 174- 187 doi: 10.1002/net.22012
19	WANG W, LIANG J, LIU R, et al A robust variable selection method for sparse online regression via the elastic net penalty[J]. Mathematics, 2022, 10 (16): 2985 doi: 10.3390/math10162985
20	WANG L, SU Z, QIAO J, et al A pseudo-inverse decomposition-based self-organizing modular echo state network for time series prediction[J]. Applied Soft Computing, 2022, 116: 108317 doi: 10.1016/j.asoc.2021.108317
21	WANG X, WANG X, MA B, et al High precision error prediction algorithm based on ridge regression predictor for reversible data hiding[J]. IEEE Signal Processing Letters, 2021, 28: 1125- 1129 doi: 10.1109/LSP.2021.3080181
22	TAN Y X Dynamics analysis of Mackey-Glass model with two variable delays[J]. Mathematical Biosciences and Engineering, 2020, 17 (5): 4513- 4526 doi: 10.3934/mbe.2020249
23	FU S, FANG X, CHEN X. Hybrid enhanced echo state network for nonlinear prediction of multivariate chaotic time series [C]// Proceedings of the IEEE International Conference on Networking, Sensing and Control. Shanghai: IEEE, 2022: 1–6.
24	YU Y, WANG Y, LUO Z, et al Link prediction algorithm based on clustering coefficient and node centrality[J]. Journal of Tsinghua University, 2022, 62 (1): 98- 104
25	DONG Z, DUAN Y, ZHOU Y, et al Weight-adaptive channel pruning for CNNs based on closeness-centrality modeling[J]. Applied Intelligence, 2024, 54 (1): 201- 215 doi: 10.1007/s10489-023-05164-5

[1]	马泽超,刘小明,夏汗青,王伟强,王久增,申海涛. 基于图神经网络的路面病害态势预测方法[J]. 浙江大学学报(工学版), 2024, 58(12): 2596-2608.
[2]	王晨霖,杨洁,居文军,顾复,陈芨熙,纪杨建. 基于智能家电的短期电力负荷预测与削峰填谷优化[J]. 浙江大学学报(工学版), 2020, 54(7): 1418-1424.
[3]	汪子龙,王柱,於志文,郭斌,周兴社. 多源数据跨国人口迁移预测[J]. 浙江大学学报(工学版), 2019, 53(9): 1759-1767.
[4]	丁理杰曹一家刘美君. 复杂电力网络的连锁故障动态模型与分析[J]. J4, 2008, 42(4): 641-646.
[5]	王磊戴华平. 基于主方程的加速生长网络分析[J]. J4, 2007, 41(2): 198-201.

Viewed

Full text

Abstract

Cited

Shared

Discussed