Supervised and semi-supervised twin parametric-margin regularized extreme learning machine
- PDF / 4,764,336 Bytes
- 24 Pages / 595.276 x 790.866 pts Page_size
- 42 Downloads / 183 Views
THEORETICAL ADVANCES
Supervised and semi‑supervised twin parametric‑margin regularized extreme learning machine Jun Ma1 Received: 20 December 2018 / Accepted: 30 March 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Twin extreme learning machine (TELM) has attracted considerable attention and achieved great success in the machine learning field. However, its performance will be severely affected when outliers exist in the dataset since TELM does not consider heteroscedasticity in practical applications. To improve the performance of TELM, a novel learning framework called twin parametric-margin extreme learning machine (TPMELM) was proposed. Further, to enhance the classification performance of our TPMELM in a semi-supervised learning setting, a Laplacian TPMELM (Lap-TPMELM) was developed by introducing manifold regularization into TPMELM. Using the geometric information of the marginal distribution embedded in unlabeled samples, Lap-TPMELM can effectively construct a more reasonable classifier. The TPMELM and Lap-TPMELM are suitable for many situations, especially when the data has heteroscedastic error structure. Moreover, the TPMELM and Lap-TPMELM are helpful in clarifying theoretical interpretation of parameters which control the bounds on proportions of support vectors and boundary errors. An efficient technique (successive over-relaxation, SOR) is applied in TPMELM and Lap-TPMELM, respectively. Experimental results show the effectiveness and reliability of the proposed methods. Keywords Extreme learning machine · Semi-supervised learning · Laplacian twin extreme learning machine · Manifold regularization · Successive over-relaxation technology
1 Introduction As an excellent machine learning tool, extreme learning machines (ELMs) [1–3] have been widely applied to classification and regression problems. Due to the simple structure, low computational cost and good versatility of ELM, in recent years, many researchers have made important contributions to ELM theory and application from different fields [4–12]. In addition, ELM provides a unified learning framework for problems such as regression, binary and multi-class classification [11]. It is well-known that support vector machines (SVMs) [12] are an effective machine learning algorithm with many advantages. However, SVMs have the challenge of high computational complexity. Recently, Jayadeva et al. [13] proposed a novel support vector machine called twin support vector machine (TSVM). Experimental results show * Ma Jun [email protected] 1
College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China
that the performance of TSVM is better than the SVMs. Due to its superior performance, many TSVM variants have been proposed in recent years [14–19]. Inspired by TSVM, Wan et al. [20] propose twin extreme learning machine (TELM) for classification problem. Although ELMs and SVMs show good generalization performance, how to select and utilize parameters is still a challenging problem. To
Data Loading...
