https://doi.org/10.1007/s10489-022-03492-6 Robust kernel ensemble regression in diversified kernel space with shared parameters Zhi-feng Liu 1 · Liu Chen 1 · Sumet Mehta 1,2 · Xiang-Jun Shen 1 · Yu-bao Cui 3 Accepted: 10 March 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022 Abstract Kernel regression is an effective non-parametric regression method. However, such regression methods have problems in choosing an appropriate kernel and its parameters. In this paper, we propose a robust kernel ensemble regression model (RKER) in diversified multiple Reproducing Kernel Hilbert Spaces (RKHSs). Motivated by multi-view data processing, we consider a kernel representation as one view of data and apply this multi-view modeling idea into the kernel regression scenario. The proposed RKER uses an ensemble idea to combine multiple individual regressors into one, where each kernel regressor is associated with a weight that is learned directly from one view of data without manual intervention. Thus, the problem of selecting kernel and its parameter in traditional kernel regression methods is overcome by finding best kernel combinations in diversified multiple solution spaces. With this multi-view modeling, RKER results in a superior overall regression performance and more robust in parameter selection. Further, we can learn the parameters in multiple RKHSs with individual specific and shared structures. Experimental results on Abalone and FaceBook datasets demonstrate that our proposed RKER model shows best performance among other state-of-the-art regression and ensemble methods, such as Random Forest, Gradient Boosting Regressor and eXtreme Gradient Boosting. Keywords Kernel regression · Ensemble regression · Multiple kernels · Shared parameters 1 Introduction Recently, kernel regression method is a popular non- parametric estimation method, which is widely used in various regression learning tasks due to its good performance in solving non-linear relationship. Such as Moreno-Salinas et.al [1] used kernel ridge regression confidence machine to identify and predict the real ship behavior with high accuracy. Also, Yang et al. [2] proposed  Xiang-Jun Shen xjshen@ujs.edu.cn  Yu-bao Cui ybbcui1975@hotmail.com; ybcui1975@njmu.edu.cn 1 School of Computer Science and Communication Engineer- ing, JiangSu University, Zhenjiang, 212013 China 2 Department of Electronics and Communication Engineering, JCDM College of Engineering, Haryana 125055, India 3 Clinical Research Center, The Affiliated Wuxi People’s Hospital of Nanjing Medical University, No. 299 at Qingyang Road, Wuxi 214023, People’s Republic of China a kernel regression method based on ridgelet theory and kernel machine, which has low computational complexity and robustness to noise. Challenges However, the performance of kernel based regression methods mainly depends on the choice of an appropriate kernel function and it is very difficult to determine an optimal kernel in practice. In order to solve the selection problem of kernel function and its parameters, Peter et al. [3] used a set of flexible nonlinear prediction functions to study the selection of kernel function and its parameters in kernel ridge regression. Samah et al. [4] proposed a technique to eliminate false edges from binary edge images by using local adaptive regression kernel as the descriptor of edge detection. Salhov et al. [5] designed the kernel by approximating the similarity between the ensemble parameters shared by multiple feature subsets. Meanwhile, some researches apply ensemble learning (EL) to kernel regression task to improve performance. For example, Berikov et al. [6] apply the concept of kernel ensemble scheme to solve a semi-supervised regression problem. They proved that the probability that an error is significantly converges to its minimum possible value as the / Published online: 25 April 2022 Applied Intelligence (2023) 53:1051–1067