计算机科学 ›› 2017, Vol. 44 ›› Issue (3): 264-267.doi: 10.11896/j.issn.1002-137X.2017.03.054

• 人工智能 • 上一篇    下一篇

一种基于CRO的高阶神经网络多示例学习方法

邓波,陆颖隽,王如志   

  1. 邵阳学院信息工程系 邵阳422000,武汉大学信息管理学院 武汉430072,北京工业大学材料科学与工程学院 北京100124
  • 出版日期:2018-11-13 发布日期:2018-11-13
  • 基金资助:
    本文受国家自然科学基金项目(61373132),湖南省教育厅科研基金资助

Multiple-instance Learning Method Based on CRO High Order Neural Networks

DENG Bo, LU Ying-jun and WANG Ru-zhi   

  • Online:2018-11-13 Published:2018-11-13

摘要: 在多示例学习(MIL)中,包是含有多个示例的集合,训练样本只给出包的标记,而没有给出单个示例的标记。提出一种基于示例标记强度的MIL方法(ILI-MIL),其允许示例标记强度为任何实数。考虑到基于梯度训练神经网络方法的计算复杂性和ILI-MIL目标函数的复杂性,利用基于化学反应优化的高阶神经网络来实现ILI-MIL,学习方法具有较强的非线性表达能力和较高的计算效率。实验结果表明,该算法比已有算法具有更加有效的分类能力,且适应范围更广。

关键词: 多示例学习,化学反应优化,高阶神经网络,分类器

Abstract: Multi-instance learning (MIL) is a variant of inductive machine learning developed recently,in which each learning example contains a bag of instances instead of a single feature vector.In this paper,we presented a novel MIL method based on the concept of instance label intensity(ILI) called ILI-MIL.Considering the complexity of the object function and the complexity of the gradient descent based training method in neural networks,we used a chemical reaction optimization (CRO) algorithm for training a high-order neural network (HONN) to implement the presented ILI-MIL method,which has more powerful nonlinear fitting capacity and high computation efficiency.The experiment results show that our ILI-MIL method have more effective ability of classification than the state-of-the-art methods.

Key words: MIL,CRO,HONN,Classifier

[1] DIETTERICH T G,LATHROP R H,L OZANO-PEREZ T.Solving the multiple instance problem with axis-parallel rectangles[J].Articial Intelligence,1997,9(1/2):31-71.
[2] ANDRDWS S,TSOCHANTARIDIS I,H OFMANN T.Support vector machines for multiple-instance learning[M]∥Advances in Neural Information Processing Systems.2002:561-568.
[3] XU Y Y,SHIH C H.Image Classification via Multiple-Instance Decision-Based Neural Networks[C]∥2013 Seventh International Conference on Image and Graphics (ICIG).IEEE,2013:394-399.
[4] TARRAG D S,CORNELIS C,BELLO R,et al.A multi-instance learning wrapper based on the Rocchio classifier for web index recommendation[J].Knowledge-Based Systems,2014,59:173-181.
[5] FOULDS J,FRANK E.A review of multi-instance learning assumptions[J].The Knowledge Engineering Review,2010,25(1):1-25.
[6] RAMON J,DE RAEDT L.Multi instance neural networks[C]∥Proceedings of the ICML-2000 Workshop on Attribute-value and Relational Learning.2000:53-60.
[7] ZHOU Z H,ZHANG M L.Neural networks for multi-instance learning[C]∥Proceedings of the International Conference on Intelligent Information Technology.Beijing,China,2002:455-459.
[8] XU Y Y,SHIH C H.Multiple-instance learning via decision based neural networks[M]∥Intelligent Decision Technologies.Springer Berlin Heidelberg,2011:885-895.
[9] GHOSH J,SHIN Y.Efficient higher-order neural networks forclassification and function approximation[J].Int J Neural Syst,1992,3:323-350.
[10] LAM A Y S,LI V O K.Chemical-reaction-inspired metaheuristic for optimization[J].IEEE Transactions on Evolutionary Computation,2010,14(3):381-399.
[11] LAM A Y S,LI V O K,XU J.On the convergence of chemical reaction optimization for combinatorial optimization[J].IEEE Transactions on Evolutionary Computation,2013,17(5):605-620.

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!