A Threshold Accepting Approach for High Breakdown Estimation
High breakdown estimation allows one to get reasonable estimates of the parameters from a sample of data even if that sample is contaminated by large numbers of awkwardly placed outliers. A drawback of high breakdown estimation techniques is the increased computational effort often needed compared to nonhigh breakdown estimation methods. Robust estimators possessing the exact fit property, for example, are NP-hard to compute. There is no hope to compute exact solutions for large high dimensional data sets. To tackle this problem, search heuristics are used to compute NP-hard estimators in high dimensions. Here, a threshold accepting algorithm that is applicable, to different high breakdown estimators is presented. Further, variants of this threshold accepting algorithm for selected estimators—most prominently least trimmed squares and least median of squares—are introduced and shown to outperform existing popular search heuristics in difficult data situations. The results increase the applicability of robust methods and underline the usefulness of heuristics computation for computational statistics.
Threshold accepting algorithm high breakdown estimation robust regression least trimmed squares (LTS) least quantile of squares (LQS)
Huirong Cao Fuchang Wang
College of Mathematics and Information Science Langfang Teachers College Langfang, China Department of Basic Courses Institute of Disaster Prevention of CEA Sanhe, China
国际会议
成都
英文
531-534
2010-12-17（万方平台首次上网日期，不代表论文的发表时间）