Title: A probabilistic kernel approach for solving the multi-instance learning problems with different assumptions

Authors: Lixin Shen; Jianjun He; Shuang Qiao

Addresses: Transportation Management College, Dalian Maritime University, Dalian, 116026, China ' College of Information and Communication Engineering, Dalian Nationalities University, Dalian, 116600, China; Faculty of Electronic Information and Electrical Engineering, Dalian University of Technology, Dalian, 116024, China ' School of Management, Dalian Polytechnic University, Dalian, 116034, China

Abstract: Multi-instance learning (MIL) has received more and more attention in the machine learning research field due to its theoretical interest and its applicability to diverse real-world problems. In this paper, we present a probabilistic kernel approach for the multi-instance learning problems with various multi-instance assumptions by imposing Gaussian process prior on an unobservable latent function defined on the instance space. Because the relationship between the bag and its instances, triggered by the multi-instance assumption, can be exactly captured by defining the likelihood function, we can deal with different multi-instance assumptions by employing different likelihood functions. Experimental results on several multi-instance problems show that the proposed algorithms are valid and can achieve superior performance to the published MIL algorithms.

Keywords: multi-instance learning; MIL; probabilistic kernel; image categorisation; aggregate function; machine learning.

DOI: 10.1504/IJAOM.2013.055881

International Journal of Advanced Operations Management, 2013 Vol.5 No.3, pp.282 - 298

Received: 21 Jan 2013
Accepted: 29 Mar 2013
Published online: 28 Apr 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article