Review of Some Aspects of Robust Inference for Time Series.

摘要

This paper briefly surveys some aspects of robust inference for time series, and gives an indication of the current state of knowledge in other problem areas. Basic notions of robustness are stated, and technical difficulties associated with the time series case are mentioned. Some models for time series with outliers are given. Least-squares procedures lack robustness for such models and robust alternatives are described. Issues of adaptivity versus robustness are briefly mentioned. Robustness problems involving dependency are discussed. Algorithms for robust data smoother-cleaners are briefly described, along with an application to radar glint noise.