Predicting invasive species distributions using incremental ensemble-based pseudo-labeling

Eunbeen Kim, Jaeuk Moon, Jonghwa Shim, Eenjun Hwang

Abstract: Highlights•An invasive species distribution prediction scheme was proposed to solve an insufficient labeled data problem.•A novel ensembled-based pseudo-labeling approach was proposed for producing accurate pseudo-labels.•A data selection technique was suggested to utilize confident pseudo-labels.•The effectiveness of the proposed scheme was evaluated using citizen science datasets of three species.•The proposed scheme outperformed other popular species distribution prediction models in terms of discrimination performance.