G-mean based extreme learning machine for imbalance learning
| Autor: | Jong-Hyok Ri, Hun Kim |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Optimization problem
Generalization Computer science 02 engineering and technology Machine learning computer.software_genre Convolutional neural network Artificial Intelligence 0202 electrical engineering electronic engineering information engineering Feature (machine learning) Electrical and Electronic Engineering Representation (mathematics) Extreme learning machine business.industry Applied Mathematics 020206 networking & telecommunications Computational Theory and Mathematics Signal Processing Metric (mathematics) 020201 artificial intelligence & image processing Computer Vision and Pattern Recognition Artificial intelligence Statistics Probability and Uncertainty business Feature learning computer |
| Zdroj: | Digital Signal Processing. 98:102637 |
| ISSN: | 1051-2004 |
| DOI: | 10.1016/j.dsp.2019.102637 |
| Popis: | Although extreme learning machine (ELM) provides better generalization performance at a much faster learning speed than traditional learning algorithms, the classical ELM can not obtain ideal results for the imbalanced data problem. In this paper, in order to conquer the learning capability of the classical ELM for an imbalance data learning, we define a new cost function of ELM optimization problem based on G-mean widely used as evaluation metric in imbalance data learning. We perform experiments on standard classification datasets which consist of 58 binary datasets and 11 multi-class datasets with the different degrees of the imbalance ratio. Experimental results show that proposed algorithm can improve the classification performance significantly compared with other state-of-the-art methods. We also demonstrate that our proposed algorithm can achieve high accuracy in representation learning by performing experiments on YouTube-8M with feature representation from convolutional neural networks. Statistical results indicate that the proposed approach not only outperforms the classical ELM, but also yields better or at least competitive results compared with several state-of-the-art class imbalance learning approaches. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect