Meta-feature selection method based on ant lion optimization algorithm

被引：0

作者：

Li G. ^{[1
]}

Liu Y. ^{[1
]}

Zheng Q. ^{[2
]}

Qin W. ^{[1
]}

Li H. ^{[2
]}

Ren X. ^{[1
]}

Song M. ^{[3
]}

机构：

[1] Defense Innovation Institute, Beijing

[2] Academy of Military Science, Beijing

[3] Tianjin (Binhai) Artificial Intelligence Innovation Center, Tianjin

来源：

Xi Tong Gong Cheng Yu Dian Zi Ji Shu/Systems Engineering and Electronics | 2023年 / 45卷 / 09期

关键词：

algorithm selection; ant lion optimization (ALO) algorithm; classification; meta-feature selection; meta-learning;

D O I：

10.12305/j.issn.1001-506X.2023.09.22

中图分类号：

学科分类号：

摘要：

To improve the performance of meta-learning algorithm selection, a meta-feature selection method based on ant lion algorithm is proposed. Firstly, an initial population is constructed through a robust initialization mechanism to enhance the robustness of the selected meta-feature subset. Secondly, dynamic boundary strategy is applied in the search process of individual solutions to increase the population diversity of the method. Then, the chaos map mutation strategy is used to improve the optimization performance of the method, give the method pseudocode and analyze the time complexity. Finally, classification algorithm selection problem are constructed using 130 datasets, 150 meta features, eight candidate algorithms, and five performance indicators for testing experiments. The parameter sensitivity and mechanism strategy effectiveness of the method are analyzed. The performance of the method is evaluated and compared through accuracy, precision, recall, and F1 score indicators, verifying the effectiveness and superiority of the proposed method. © 2023 Chinese Institute of Electronics. All rights reserved.

引用

下载

页码：2831 / 2842

页数：11

共 34 条

[1] ADAMS P, ALEXANDROPOULOS S-A N, PARDALOS P M, Et al., No free lunch theorem
[2] a review, Approximation and optimization, pp. 57-82, (2019)
[3] KERSCIIKEP, IIOOSIIII, NEUMANN F, Et al., Automated algorithm selection: survey and perspectives [J], Evolutionary Computation, 27, 1, pp. 3-45, (2019)
[4] BRAZDIL P, GIRAUD-CARRIER C., Metalearning and algorithm selection: progress, state of the art and introduction to the 2018 Special Issue, Machine Learning, 107, 1, pp. 1-14, (2018)
[5] YANG C, AKIMOTO Y, KIM D W, Et al., OBOE: collaborative filtering for AutoML model selection, Proc. of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 1173-1183, (2019)
[6] LI R F, XU A Q, SUN W C, Et al., Recommendation method for avionics feature selection algorithm based on meta-learning[J], Systems Engineering and Electronics, 43, 7, pp. 2011-2020, (2021)
[7] DIAS L V, MIRANDA P B C, NASCIMENTO A C A, Et al., ImageDataset2Vec: an image dataset embedding for algorithm selection [ J ], Expert Systems with Applications, 180, (2021)
[8] KOTLAR M, PUNT M, RADIVOJEVIC Z, Et al., Novel meta-features for automated machine learning model selection in anomaly detectionDQ, IEEE Access, 9, pp. 89675-89687, (2021)
[9] RIVOLLI A, GARCIA L P F, SOARES C, Et al., Meta-features for meta-learning, Knowledge-Based Systems, 2, 240
[10] GARCIA-SAIZ D, ZORRILLA M., A meta-learning based framework for building algorithm recommenders: an application for educational arena [J], Journal of Intelligent & Fuzzy Systems, 32, 2, pp. 1449-1459, (2017)

← 1 2 3 4 →