Abstract:Least squares support vector machine for regression (LSSVR) is an efficient method for function estimation problem. However, its solution is prone to large noise and outliers since it depends on the minimum of the sum of squares error (SSE) on training samples. To tackle this problem, in this paper, a novel regression model termed as recursive robust LSSVR (R^2LSSVR) is proposed to obtain robust estimation for data in the presence of outliers. The idea is to build a regression model in the kernel space based on maximum correntropy criterion and regularization technique. An iterative algorithm derived from half-quadratic optimization is further developed to solve R^2LSSVR with theoretically guaranteed convergence. It also reveals that R^2LSSVR is closely related to the original LSSVR since it essentially solves adaptive weighted LSSVR iteratively. Furthermore, a hyperparameters selection method for R^2LSSVR is presented based on particle swarm optimization (PSO) such that multiple hyperparameters in R^2LSSVR can be estimated effectively for better performance. The feasibility of this method is examined on some simulated and benchmark datasets. The experimental results demonstrate the good robust performance of the proposed method.

Theoretic Research on Robustified Least Squares Estimator Based on Equivalent Variance-Covariance

Robust estimation of variance components model

A compromise criterion for weighted least squares estimates

Equivalent Weight Robust Estimation Method Based on Median Parameter Estimates

Robust Estimate with Minimum Mean Squared Error

A New Robust Estimation Method and Its Calculation

APPLICATION OF L-CURVE METHOD TO EQUIVALENT WEIGHT ROBUST RIDGE ESTIMAION MODEL

Robust M-estimation Using the Equivalent Weights Constructed by Removing the Influence of an Outlier on the Residuals

DEPENDENT ROBUST ESTIMATION AND ITS APPLICATIONS IN SURVEY DATA PROCESSING

Robustness and the Robust Estimate

Variance-covariance Component Estimation Based on the Equivalent Residuals

Recursive Robust Least Squares Support Vector Regression Based on Maximum Correntropy Criterion

Sparse Estimation Via ℓ_q Optimization Method in High-Dimensional Linear Regression

Kernel-based Regression Via a Novel Robust Loss Function and Iteratively Reweighted Least Squares

A universal formula of robust estimation of variance component

Weighted least squares regression with the best robustness and high computability

The Robust Kalman Filtering with Continuous Variable Equivalent Weight Function

Weighted Stochastic Restricted Estimation in Linear Measurement Error Models

RCR: Robust Compound Regression for Robust Estimation of Errors-in-Variables Model

Robust MAVE for single-index varying-coefficient models

Variance Analysis of Robust State Estimation in Power System Using Influence Function