基于群智能增强核极限学习机的创新人才预测模型

doi:10.13229/j.cnki.jdxbgxb.20240906

摘要/Abstract

摘要：

为解决高校创新人才传统预测方法存在主观性强、准确度低等问题，提出了一种融合信息引导交流搜索策略的粒子群优化算法与核极限学习机的高校创新人才智能预测模型。该模型利用改进的粒子群优化算法增强种群多样性与全局寻优能力，以提升核极限学习机的分类性能，旨在更科学、客观地识别与选拔创新人才。为验证模型有效性，通过十折交叉验证在高校创新人才数据集上进行实验，结果表明：本文模型在分类准确率（86.05%）、灵敏度（89.74%）、特异性（83.24%）和马修斯相关系数（72.42%）上均优于多种对比模型。研究结果证实了该模型在高校创新人才预测方面的显著优势，为人才的科学选拔与培养提供了新的技术手段，具备良好的应用前景。

关键词: 机器学习, 创新人才预测, 核极限学习机, 粒子群优化

Abstract:

To address the issues of strong subjectivity and low accuracy in traditional methods for predicting innovative talents in higher education， this paper proposes an intelligent predictive model that combines a Particle Swarm Optimization algorithm， enhanced with an Information-Guided Communication Search strategy， with a Kernel Extreme Learning Machine. This model aims to more scientifically and objectively identify and select innovative talents by utilizing the improved Particle Swarm Optimization algorithm to enhance population diversity and global search capabilities， thereby improving the classification performance of the Kernel Extreme Learning Machine. To validate its effectiveness， experiments were conducted on a university innovation talent dataset using 10-fold cross-validation. The results demonstrate that the proposed model outperforms several comparative models in key evaluation metrics， including classification accuracy （86.05%）， sensitivity （89.74%）， specificity （83.24%）， and Matthews correlation coefficient （72.42%）. These findings confirm the significant advantages of the proposed model in predicting innovative university talents， offering a new technical approach for the scientific selection and cultivation of talent with promising application prospects.

Key words: machine learning, prediction of innovative talent, kernel extreme learning machine, particle swarm optimization

中图分类号:

TP391.4

金庆良,周鑫森,陈翼,吴承文. 基于群智能增强核极限学习机的创新人才预测模型[J]. 吉林大学学报(工学版), 2025, 55(5): 1763-1771.

Qing-liang JIN,Xin-sen ZHOU,Yi CHEN,Cheng-wen WU. Predictive model for identifying innovative university talents based on the swarm intelligence evolution enhanced kernel extreme learning machine[J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(5): 1763-1771.

图/表 8

图1

表1

特征属性描述"

标记	特征
F1	性别
F2	年级
F3	专业
F4	您认为自己主要成长于哪方面
F5	生源地
F6	对目前所学的专业感兴趣
F7	每周课外花在专业学习上的时间有多少
F8	您会主动寻找与本专业相关的课外资源和实践机会吗
F9	学习专业知识感到开心
F10	所学专业课程内容对个人成长和未来发展有多大价值
F11	完成专业课程项目或任务时，能感受到明显的成就感
F12	积极参与课程相关的实践活动，并有所收获
F13	面对一些任务时，经常只求能应付过去
F14	常常考虑仔细后才作出决定
F15	在他人眼中我是一个慎重的人
F16	做事讲究逻辑和条理是我的一个特点
F17	我擅长于将问题转化为机会
F18	我一直在寻找更好的行事方式
F19	如果我坚信某件事，不管成败的可能性如何，我都会去做
F20	没有比看到我的想法变成现实更令人兴奋的事了
F21	我是一个月光族
F22	当我在超市购物时，会买很多计划外的东西
F23	人应该及时行乐
F24	如果为了得到一件东西我必须要做出计划和等待，那么我会更加享受这件东西
F25	您父亲的最高学历
F26	您母亲的最高学历
F27	您父母有从事和科学研究相关的职业
F28	在您的童年时期，您父母看书的频率如何
F29	您家庭成员间会讨论科学新闻、科技进展或科研成果
F30	您认为您家庭的年总收入大致处于哪个区间
F31	在高中时期，您每年的教育支出约为多少
F32	您认为父母对您的学术事业支持程度如何
F33	您对当前所学专业的兴趣有多大程度受父母影响
F34	您是独生子女吗？如果不是，那么您是家里的第几个孩子
F35	您高中阶段老师对专业知识的掌握程度如何
F36	您高中阶段的老师有效地将复杂概念讲解得通俗易懂
F37	您高中阶段的老师鼓励学生进行独立思考和探索性学习
F38	您高中阶段的老师主动引导学生关注科学前沿动态，培养科研兴趣
F39	您高中阶段的老师展现出对科学的热情和敬业精神，并对您产生积极影响
F40	您获得过奥林匹克竞赛等学科竞赛的省级或省级以上奖项
F41	您在上大学前参与过科技创新类活动并完成项目
F42	您所在的高中举行过与大学对接的讲座、课程或实践活动
F43	您认为你家乡的文化传统有助于你的学业
F44	您平均每天花费多少时间浏览社交媒体（如微博、抖音、小红书、知乎等）的内容
F45	在社交媒体推送的内容中，更关注学术前沿的内容
F46	在社交媒体推送的内容中，更关注哲学思辨的内容
F47	在社交媒体推送的内容中，更关注文化艺术的内容
F48	在社交媒体推送的内容中，更关注社会热点的内容
F49	在社交媒体推送的内容中，更关注休闲娱乐的内容
F50	您认为浏览社交媒体的相关内容有助于你的学业
F51	在您的求学过程中，有受到国家有关政策的帮助和支持

表1

表2

比较算法的参数设置"

方法	参数值
bICSPSO-KELM	$c 1 = 2; c 2 = 2; V m a x = 6$
PbGSK_V4-KELM	$m i n P o p = 12; f u n_n o = 8$
bPSO-KELM	$w ∈ 0.2, ? 0.9; c 1 = c 2 = 2$
bABC-KELM	$l i m i t = 300; m = 2$
XGBoost	$η = 0.3; m a x ?_d e p t h = 6$
RandomF	$n T r e e = 20$
KELM	$c = 88; γ = 1024$
SVM	$c = 850; γ = 0.17$

表2

图2

表3

表4

图3

图4

参考文献 21

[1]	Li J, Xue E.How talent cultivation contributes to creating world-class universities in China:a policy discourse analysis[J].Educational Philosophy and Theory, 2022, 54(12): 2008-2017.
[2]	史秋衡, 李瑞. 高校拔尖创新人才培养的价值逻辑、关键要素与路径选择[J]. 中国远程教育, 2024, 44(1): 15-24.
	Shi Qiu-heng, Li Rui. Value logic, key elements and path selection of top innovative talents training in universities[J]. Distance Education in China, 2024, 44(1): 15-24.
[3]	李润亚, 张潮, 张珂, 等. 大学生创新创业能力系统构成及其表现研究[J]. 教育理论与实践, 2024, 44(18): 10-15.
	Li Run-ya, Zhang Chao, Zhang Ke, et al. A study on the system composition and performance of college students' innovation and entrepreneurship ability[J]. Educational Theory and Practice,2024,44(18):10-15.
[4]	Gao H W.Innovation and development of ideological and political education in colleges and universities in the network era[J]. International Journal of Electrical Engineering& Education,2021,60(Sup.2): 489-499.
[5]	李立国, 刘振天, 陈恩伦, 等. 笔谈:大学可持续发展的多维思考[J] .现代大学教育, 2023, 39(6): 41-59.
	Li Li-guo, Liu Zhen-tian, Chen En-lun, et al. Writing: multi-dimensional thinking on sustainable development of universities [J]. Modern University Education,2023,39(6): 41-59.
[6]	徐玉成, 王波, 朱萍. 科教融汇赋能职业教育人才培养的时代价值、现实困境及破解对策[J].教育学术月刊, 2023(9): 58-66.
	Xu Yu-cheng, Wang Bo, Zhu Ping. The times value, realistic dilemma and solutions of talent training in enabling vocational education by integrating science and education[J]. Educational Academic Monthly, 2023(9): 58-66.
[7]	Liu Y. Construction of talent training mechanism for innovation and entrepreneurship education in colleges and universities based on data fusion algorithm[J].Frontiers in Psychology,2022,13: 968023.
[8]	周沂, 陈圆月, 冯皓月. 学科交叉推动拔尖创新人才培养的作用研究——来自S大学书院制的经验证据[J]. 湖南师范大学教育科学学报, 2024, 23(1): 33-43.
	Zhou Yi, Chen Yuan-Yue, Feng Hao-yue. Research on the role of interdisciplinary interaction in promoting the cultivation of top innovative talents: empirical evidence from the college system of S University[J]. Journal of Education Science, Hunan Normal University, 2024, 23(1): 33-43.
[9]	Sarker I H.Machine learning:algorithms, real-world applications and research directions[J]. SN Computer Science, 2021, 2(3): No.160.
[10]	Mishra D K, Shinde V.6 A review of global optimization problems using meta-heuristic algorithm[M]∥Aditya K, Ashish K, Nhu Gia N, et al. Nature-Inspired Optimization Algorithms.Berlin,Boston;De Gruyter, 2021: 87-106.
[11]	Storn R, Price K. Differential evolution—a simple and efficient heuristic for global optimization over continuous spaces[J]. Journal of Global Optimization,1997, 11(4): 341-359.
[12]	Kennedy J, Eberhart R.Particle swarm optimization[C]∥Proceedings of ICNN'95-International Conference on Neural Networks, Perth, Australia, 1995:488968.
[13]	Dorigo M, Birattari M, Stutzle T. Ant colony optimization[J]. IEEE Computational Intelligence Magazine, 2006, 1(4): 28-39.
[14]	Mirjalili S, Mirjalili S M, Lewis A. Grey wolf optimizer[J].Advances in Engineering Software, 2014, 69: 46-61.
[15]	Lian J, Hui G, Ma L, et al.Parrot optimizer: algorithm and applications to medical problems[J]. Computers in Biology and Medicine, 2024, 172: 108064.
[16]	Chauhan S, Vashishtha G, Abualigah L, et al.Boosting salp swarm algorithm by opposition-based learning concept and sine cosine algorithm for engineering design problems[J].Soft Computing, 2023, 27(24):18775-18802.
[17]	Gao R, Tao J, Zhang J,et al.NSGA-Ⅲ-SD based fuzzy energy management system optimization for lithium battery/supercapacitor HEV[J].Applied Soft Computing, 2023, 142: 110280.
[18]	Dong R, Sun L, Ma L, et al.Boosting kernel search optimizer with slime mould foraging behavior for combined economic emission dispatch problems[J]. Journal of Bionic Engineering, 2023, 20(6): 2863-2895.
[19]	Pan X, Zhang G, Lin A,et al.An evaluation model for children's foot& ankle deformity severity using sparse multi-objective feature selection algorithm[J]. Computers in Biology and Medicine, 2022, 151: 106229.
[20]	Huang G B, Zhou H, Ding X,et al. Extreme learning machine for regression and multiclass classification[J].IEEE Transactions on Systems, Man, and Cybernetics,Part B(Cybernetics), 2012, 42(2): 513-529.
[21]	Huang G B, Zhu Q Y, Siew C K. Extreme learning machine: theory and applications[J]. Neurocomputing, 2006, 70(1): 489-501.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

转换函数	bICSPSO-KELM
	Accuracy		Sensitivity		Specificity		MCC
	均值/%	方差	均值/%	方差	均值/%	方差	均值/%	方差
S1	85.02	0.03	87.94	0.04	83.13	0.05	70.45	0.06
S2	85.19	0.04	88.85	0.04	82.54	0.04	70.79	0.07
S3	85.00	0.04	86.80	0.06	83.91	0.05	70.32	0.08
S4	85.10	0.03	88.77	0.03	82.43	0.04	70.59	0.07
V1	86.05	0.04	89.74	0.03	83.24	0.05	72.42	0.08
V2	85.67	0.02	89.28	0.03	83.24	0.04	71.84	0.03
V3	85.96	0.03	88.97	0.03	83.61	0.04	72.16	0.06
V4	85.67	0.02	90.82	0.05	82.38	0.04	72.11	0.05

折数	Accuracy	Sensitivity	Specificity	MCC
#1	84.62%	88.89%	81.36%	69.62%
#2	86.67%	88.00%	85.46%	73.38%
#3	91.43%	93.75%	89.47%	82.95%
#4	81.91%	86.67%	78.33%	64.34%
#5	91.43%	93.88%	89.29%	82.98%
#6	79.81%	84.09%	76.67%	60.05%
#7	90.48%	93.62%	87.93%	81.13%
#8	83.81%	88.64%	80.33%	68.08%
#9	86.54%	89.36%	84.21%	73.25%
#10	83.81%	90.48%	79.37%	68.46%
均值	86.05%	89.74%	83.24%	72.42%
方差	0.04	0.03	0.05	0.08