.png){kind=icon}
返回
Boost two-view learning-based method for label proportions problem
.png){kind=icon}
PRE
AI
DOI：10.1007/s10489-023-04643-z.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
In this paper, we mainly research the problem of learning from label proportions (LLP), in which the training data is divided into several bags, and only the label proportions information of each class can be accessible. As it has drawn increasing attention recently, there are a great many approaches have been proposed to solve the problem of LLP, while most existing LLP methods only consider a single-view feature of the training data, in which the classifiers are more likely to obtain deficient performance and be interfered by noise. Thus, we proposed a novel approach to solve the problem of LLP based on two-view learning, term two-view learning from label proportions (TV-LLP). The proposed method can build a more satisfying classifier by combining two-view knowledge than traditional LLP methods with a single view. In the first place, we formulate the TV-LLP model to deal with two-view learning in the label proportions setting and solve it iteratively, in which we assign weight distribution to each classier for acquiring basic classifiers. Meanwhile, we introduce boosting method into the model to construct a strong classifier based on the basic classifiers, which can improve the accuracy of the model further. We then apply the proposed method to the datasets of text categorization and image categorization respectively, in which extensive experiments have shown that the proposed TV-LLP method outperforms existing approaches.
Keyword:
Label proportions problem
Two-view learning
Boosting
期刊
IF:
3.5
论文数:
7.6K
被引数:
1.7W
机构
引用论文
Detection of Leishmania infantum DNA in the non-parasitized lung of dogs with visceral leishmaniasis
OAAI
.png){kind=icon}
OAAI分享收藏Ecología urbana de Triatoma infestans en Argentina: asociación entre Triatoma infestans y palomaresOAAI
OAAI分享收藏Crystalline‐State Reaction with Allosteric Effect in Spin‐Crossover, Interpenetrated Networks with Magnetic and Optical Bistability具有磁和光学双稳态的自旋交叉，互穿网络中具有变构效应的晶态反应
PREAI
分享收藏A practical tutorial on the use of nonparametric statistical tests as a methodology for comparing evolutionary and swarm intelligence algorithms关于使用非参数统计检验作为比较进化和群体智能算法的方法的实用教程
PREAI
分享收藏
查看更多内容