Robust Relevance Vector Regression With Trimmed Likelihood Function
Biao Yang, Zengke Zhang, Zhengshun Sun · IEEE Signal Processing Letters · 2007
This letter proposes a novel robust regression method called trimmed relevance vector regression (TRVR) that redefines the likelihood function as a trimmed likelihood function over a trimmed subset. A re-weighted strategy is introduced to find the robust trimmed subset that does not include outliers. Simultaneously, by maximizing the trimmed likelihood function with the relevance vector machine (RVM) framework, the weights of the regression model can be estimated. The experimental results have been presented to demonstrate that the proposed method is highly robust.