Journal of Computer Applications ›› 2016, Vol. 36 ›› Issue (2): 387-391.DOI: 10.11772/j.issn.1001-9081.2016.02.0387

Previous Articles     Next Articles

Semi-supervised extreme learning machine and its application in analysis of near-infrared spectroscopy data

JING Shibo1, YANG Liming1, LI Junhui2, ZHANG Siyun1   

  1. 1. College of Science, China Agricultural University, Beijing 100083, China;
    2. College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China
  • Received:2015-08-29 Revised:2015-09-15 Online:2016-02-10 Published:2016-02-03

半监督极限学习机及其在近红外光谱数据分析中的应用

井诗博1, 杨丽明1, 李军会2, 张思韫1   

  1. 1. 中国农业大学 理学院, 北京 100083;
    2. 中国农业大学 信息与电气工程学院, 北京 100083
  • 通讯作者: 杨丽明(1963-),女,河北保定人,副教授,博士,主要研究方向:数据挖掘、机器学习、非线性规划。
  • 作者简介:井诗博(1992-),女,吉林长春人,硕士研究生,主要研究方向:数据挖掘、机器学习;李军会(1975-),男,山东莱芜人,副教授,博士,主要研究方向:光谱信号处理;张思韫(1993-),女,山西晋城人,主要研究方向:数据挖掘、机器学习。
  • 基金资助:
    国家自然科学基金资助项目(11471010,11271367)。

Abstract: When insufficient training information is available, supervised Extreme Learning Machine (ELM) is difficult to use. Thus applying semi-supervised learning to ELM, a Semi-Supervised ELM (SSELM) framework was proposed. However, it is difficult to find the optimal solution of SSELM due to its nonconvexity and nonsmoothness. Using combinatorial optimization method, SSELM was solved by reformulating SSELM as a linear mixed integer program. Furthermore, SSELM was used for the direct recognition of medicine and seeds datasets using Near-InfraRed spectroscopy (NIR) technology. Compared with the traditional ELM methods, the experimental results show that SSELM can improve the generation when insufficient training information is available, which indicates the feasibility and effectiveness of the proposed method.

Key words: Extreme Learning Machine(ELM), semi-supervised learning, nonconvex optimization, mixed integer programming, Near-InfraRed spectroscopy(NIR)

摘要: 当数据集中包含的训练信息不充分时,监督的极限学习机较难应用,因此将半监督学习应用到极限学习机,提出一种半监督极限学习机分类模型;但其模型是非凸、非光滑的,很难直接求其全局最优解。为此利用组合优化方法,将提出的半监督极限学习机化为线性混合整数规划,可直接得到其全局最优解。进一步,利用近红外光谱技术,将半监督极限学习机应用于药品和杂交种子的近红外光谱数据的模式分类。与传统方法相比,在不同的光谱区域的数值实验结果显示:当数据集中包含训练信息不充分时,提出的半监督极限学习机提高了模型的推广能力,验证了所提出方法的可行性和有效性。

关键词: 极限学习机, 半监督学习, 非凸最优化, 混合整数规划, 近红外光谱

CLC Number: