
Research Article
A Semi-supervised Learning Application for Hand Posture Classification
@INPROCEEDINGS{10.1007/978-3-031-33614-0_10, author={Kailiang Nan and Shengnan Hu and Haozhe Luo and Patricia Wong and Saeid Pourroostaei Ardakani}, title={A Semi-supervised Learning Application for Hand Posture Classification}, proceedings={Big Data Technologies and Applications. 11th and 12th EAI International Conference, BDTA 2021 and BDTA 2022, Virtual Event, December 2021 and 2022, Proceedings}, proceedings_a={BDTA}, year={2023}, month={5}, keywords={Semi-supervised learning Big data Analysis Co-forest Tri-training hand posture classification}, doi={10.1007/978-3-031-33614-0_10} }
- Kailiang Nan
Shengnan Hu
Haozhe Luo
Patricia Wong
Saeid Pourroostaei Ardakani
Year: 2023
A Semi-supervised Learning Application for Hand Posture Classification
BDTA
Springer
DOI: 10.1007/978-3-031-33614-0_10
Abstract
The rapid growth of HCI applications results in increased data size and complexity. For this, advanced machine learning techniques and data analysis solutions are used to prepare and process data patterns. However, the cost of data pre-processing, labelling, and classification can be significantly increased if the dataset is huge, complex, and unlabelled. This paper aims to propose a data pre-processing approach and semi-supervised learning technique to prepare and classify a big Motion Capture Hand Postures dataset. It builds the solutions via Tri-training and Co-forest techniques and compares them to figure out the best-fitted approach for hand posture classification. According to the results, Co-forest outperforms Tri-training in terms of Accuracy, Precision, recall, and F1-score.