Label distribution learning via label correlation grid

TL;DR

This paper introduces a Label Correlation Grid (LCG) to model label relationship uncertainties in label distribution learning, improving estimation accuracy through covariance modeling and a projection regularization, validated on real benchmarks.

Contribution

The paper proposes a novel LCG approach that models label relationship uncertainty using covariance matrices and Gaussian distributions, enhancing label distribution estimation.

Findings

Effective in reducing noise in label space

Improves label distribution estimation accuracy

Validated on multiple real-world benchmarks

Abstract

Label distribution learning can characterize the polysemy of an instance through label distributions. However, some noise and uncertainty may be introduced into the label space when processing label distribution data due to artificial or environmental factors. To alleviate this problem, we propose a \textbf{L}abel \textbf{C}orrelation \textbf{G}rid (LCG) to model the uncertainty of label relationships. Specifically, we compute a covariance matrix for the label space in the training set to represent the relationships between labels, then model the information distribution (Gaussian distribution function) for each element in the covariance matrix to obtain an LCG. Finally, our network learns the LCG to accurately estimate the label distribution for each instance. In addition, we propose a label distribution projection algorithm as a regularization term in the model training process. Extensive experiments verify the effectiveness of our method on several real benchmarks.