Nested Multiple Instance Learning with Attention Mechanisms

TL;DR

This paper introduces NMIA, a nested multiple instance learning model with attention mechanisms, capable of handling complex hierarchical data and revealing latent labels at multiple levels, improving upon traditional MIL methods.

Contribution

The paper proposes a novel NMIA architecture that combines nesting and attention to better model complex hierarchical data in weakly supervised learning.

Findings

NMIA performs comparably to traditional MIL in simple scenarios.

NMIA effectively captures complex hierarchical relationships.

NMIA provides insights into latent labels at different levels.

Abstract

Strongly supervised learning requires detailed knowledge of truth labels at instance levels, and in many machine learning applications this is a major drawback. Multiple instance learning (MIL) is a popular weakly supervised learning method where truth labels are not available at instance level, but only at bag-of-instances level. However, sometimes the nature of the problem requires a more complex description, where a nested architecture of bag-of-bags at different levels can capture underlying relationships, like similar instances grouped together. Predicting the latent labels of instances or inner-bags might be as important as predicting the final bag-of-bags label but is lost in a straightforward nested setting. We propose a Nested Multiple Instance with Attention (NMIA) model architecture combining the concept of nesting with attention mechanisms. We show that NMIA performs as conventional MIL in simple scenarios and can grasp a complex scenario providing insights to the latent labels at different levels.