Learning naive Bayes models for multiple-instance learning with label proportions

Authors:
Jerónimo Hernández;Iñaki Inza
Affiliations:
Intelligent Systems Group, University of the Basque Country;Intelligent Systems Group, University of the Basque Country
Venue:
CAEPIA'11 Proceedings of the 14th international conference on Advances in artificial intelligence: spanish association for artificial intelligence
Year:
2011

Citing 5
Cited 1

Solving the multiple instance problem with axis-parallel rectangles

Artificial Intelligence
Estimating labels from label proportions

Proceedings of the 25th international conference on Machine learning
Supervised Learning by Training on Aggregate Outputs

ICDM '07 Proceedings of the 2007 Seventh IEEE International Conference on Data Mining
Kernel K-means Based Framework for Aggregate Outputs Classification

ICDMW '09 Proceedings of the 2009 IEEE International Conference on Data Mining Workshops
Estimating Labels from Label Proportions

The Journal of Machine Learning Research

Learning Bayesian network classifiers from label proportions

Pattern Recognition

Quantified Score

Hi-index	0.01

Visualization

Abstract

This paper deals with the problem of multi-instance learning when label proportions are provided. In this classification problem, the instances of the dataset are divided into disjoint groups, where there is no certainty about the labels associated with individual samples. However, in each group the number of instances that belong to each class is known. We propose several versions of an EM-algorithm that learns naive Bayes models to deal with the exposed problem. The proposed algorithms are evaluated on synthetic and real datasets, and compared with state-of-the-art approaches. The obtained results show a competitive behaviour of our proposals.