You are here

Regularized joint density estimation for multi-instance learning

TitleRegularized joint density estimation for multi-instance learning
Publication TypeConference Paper
Year of Publication2012
AuthorsBehmardi, B., F. Briggs, X. Z. Fern, and R. Raich
Conference NameIEEE Statistical Signal Processing Workshop (SSP)
Pagination740 - 743
Date Published08/2012
Conference LocationAnn Arbor, MI
ISBN Number978-1-4673-0181-7

We present regularized multiple density estimation (MDE) using the maximum entropy (MaxEnt) framework for multi-instance datasets. In this approach, bags of instances are represented as distributions using the principle of MaxEnt. We learn basis functions which span the space of distributions for jointly regularized density estimation. The basis functions are analogous to topics in a topic model. We propose a distance metric for measuring similarities at the bag level which captures the statistical properties of each bag. We provide a convex optimization method to learn the metric and compare the results with distance based multi-instance learning algorithms, e.g., Citation-kNN and bag-level kernel SVM on two real world datasets. The results show that regularized MDE produces a comparable results in terms of accuracy with reduced computational complexity.