Spatial-prior generalized fuzziness extreme learning machine autoencoder-based active learning for hyperspectral image classification

Hyperspectral imaging has been extensively utilized in several fields, and it benefits from detailed spectral information contained in each pixel, generating a thematic map for classification to assign a unique label to each sample. However, the acquisition of labeled data for classification is expensive in terms of time and cost. Moreover, manual selection and labeling are often subjective and tend to induce redundancy into the classifier. In this paper, a spatial prior generalized fuzziness extreme learning machine autoencoder (GFELM-AE) based active learning is proposed, which contextualizes the manifold regularization to the objective of ELM-AE. Experiments on a benchmark dataset confirmed that the GFELM-AE presents competitive results compared to the state-of-the-art, leading to the improved statistical significance in terms of F1-score, precision, and recall.