计算机应用 ›› 2013, Vol. 33 ›› Issue (06): 1600-1603.DOI: 10.3724/SP.J.1087.2013.01600

• 人工智能 • 上一篇    下一篇

基于隐特征空间的极限学习机模型选择

毛文涛1,赵中堂2,贺欢欢1   

  1. 1. 河南师范大学 计算机与信息工程学院, 河南 新乡 453007
    2. 郑州航空工业管理学院 计算机科学与应用系,郑州 450015
  • 收稿日期:2012-12-05 修回日期:2013-01-16 出版日期:2013-06-01 发布日期:2013-06-05
  • 通讯作者: 毛文涛
  • 作者简介:毛文涛(1980-),男,河南新乡人,讲师,博士,主要研究方向:机器学习、最优化理论;赵中堂(1976-),男,河南郑州人,讲师,硕士,主要研究方向:多媒体信息处理;贺欢欢(1991-),女,河南郑州人,主要研究方向:最优化理论、模式识别。
  • 基金资助:

    国家自然科学基金资助项目(51175135);河南省基础与前沿技术研究计划项目(122300410111);河南省重点科技攻关项目(102102210176)

Model selection of extreme learning machine based on latent feature space

MAO Wentao1,ZHAO Zhongtang2,HE Huanhuan1   

  1. 1. College of Computer and Information Engineering, Henan Normal University, Xinxiang Henan 453007, China
    2. Department of Computer Science and Application, Zhengzhou Institute of Aeronautical Industry Management, Zhengzhou Henan 450015, China
  • Received:2012-12-05 Revised:2013-01-16 Online:2013-06-05 Published:2013-06-01
  • Contact: MAO Wentao

摘要: 针对极限学习机(ELM)中冗余的隐神经元会削弱模型泛化能力的缺点,提出了一种基于隐特征空间的ELM模型选择算法。首先,为了寻找合适的ELM隐层,在ELM中添加正则项,该项为现有隐层空间到低维隐特征空间的映射函数矩阵的Frobenius范数;其次,为解决该非凸问题,采用交替优化的策略,并通过凸二次型优化学习该隐空间;最终自适应得到最优映射函数和ELM模型。分别采用UCI标准数据集和载荷识别工程数据对所提算法进行测试,结果表明,与经典ELM相比,该算法可有效提高预测精度和数值稳定性,与现有模型选择算法相比,该算法预测精度相当,但运行时间则大幅降低。

关键词: 极限学习机, 模型选择, 交替优化, 隐空间, 泛化能力

Abstract: Recently, Extreme Learning Machine (ELM) has been a promising tool in solving a wide range of classification and regression problems. However, the generalization performance of ELM will be decreased when there exits redundant hidden neurons. To solve this problem, this paper introduced a new regularizer that was the Frobenius norm of mapping matrix from hidden space to a new latent feature space. Furthermore, an alternating optimization strategy was adopted to learn the above regularization problem and the latent feature space. The proposed algorithm was tested empirically on the classical UCI data set as well as a load identification engineering data set. The experimental results show that the proposed algorithm obviously outperforms the classical ELM in terms of predictive precision and numerical stability, and needs much less computational cost than the present ELM model selection algorithm.

Key words: Extreme Learning Machine (ELM), model selection, alternating optimization, latent space, generalization ability

中图分类号: