Class Prior-Free Positive-Unlabeled Learning with Taylor Variational Loss for Hyperspectral Remote Sensing Imagery

TL;DR

This paper introduces a Taylor variational loss and a self-calibrated optimization strategy to improve positive-unlabeled learning in hyperspectral imagery, effectively balancing overfitting and underfitting with limited labeled data.

Contribution

It proposes a novel Taylor variational loss and self-calibrated optimization for PU learning in hyperspectral images, addressing overfitting issues with limited labeled data.

Findings

Effective on 7 benchmark datasets with 21 tasks

Reduces overfitting by balancing gradient contributions

Improves stability of training process

Abstract

Positive-unlabeled learning (PU learning) in hyperspectral remote sensing imagery (HSI) is aimed at learning a binary classifier from positive and unlabeled data, which has broad prospects in various earth vision applications. However, when PU learning meets limited labeled HSI, the unlabeled data may dominate the optimization process, which makes the neural networks overfit the unlabeled data. In this paper, a Taylor variational loss is proposed for HSI PU learning, which reduces the weight of the gradient of the unlabeled data by Taylor series expansion to enable the network to find a balance between overfitting and underfitting. In addition, the self-calibrated optimization strategy is designed to stabilize the training process. Experiments on 7 benchmark datasets (21 tasks in total) validate the effectiveness of the proposed method. Code is at: https://github.com/Hengwei-Zhao96/T-HOneCls.