Multiple Instance Choquet Integral Classifier Fusion and Regression for Remote Sensing Applications

TL;DR

This paper introduces novel multiple instance learning-based fusion models for classification and regression in remote sensing, capable of training with imprecise labels, demonstrating effective performance on synthetic and real-world data.

Contribution

It presents new multiple instance Choquet integral fusion models that handle ambiguous labels, advancing remote sensing data analysis methods.

Findings

Effective classification and regression performance demonstrated

Models work with imprecise, bag-level labels

Applicable to target detection and crop yield prediction

Abstract

In classifier (or regression) fusion the aim is to combine the outputs of several algorithms to boost overall performance. Standard supervised fusion algorithms often require accurate and precise training labels. However, accurate labels may be difficult to obtain in many remote sensing applications. This paper proposes novel classification and regression fusion models that can be trained given ambiguosly and imprecisely labeled training data in which training labels are associated with sets of data points (i.e., "bags") instead of individual data points (i.e., "instances") following a multiple instance learning framework. Experiments were conducted based on the proposed algorithms on both synthetic data and applications such as target detection and crop yield prediction given remote sensing data. The proposed algorithms show effective classification and regression performance.