Multi-Instance Learning by Treating Instances As Non-I.I.D. Samples

TL;DR

This paper introduces a novel multi-instance learning approach that models each bag as a graph, capturing the non-i.i.d. relationships among instances to improve learning performance.

Contribution

It proposes a graph-based multi-instance learning method that considers instance relations, unlike traditional i.i.d. assumptions, enhancing model effectiveness.

Findings

Effective in distinguishing bags using graph kernels

Outperforms traditional i.i.d.-based methods

Validated through experimental results

Abstract

Multi-instance learning attempts to learn from a training set consisting of labeled bags each containing many unlabeled instances. Previous studies typically treat the instances in the bags as independently and identically distributed. However, the instances in a bag are rarely independent, and therefore a better performance can be expected if the instances are treated in an non-i.i.d. way that exploits the relations among instances. In this paper, we propose a simple yet effective multi-instance learning method, which regards each bag as a graph and uses a specific kernel to distinguish the graphs by considering the features of the nodes as well as the features of the edges that convey some relations among instances. The effectiveness of the proposed method is validated by experiments.