Elastic arm methods of data analysis as a robust approach

Rok, strany: 2003, 291 - 306

One of common problems of experimental physics is the detection of curves, such as charged particle tracks or Cherenkov rings, within a complex pattern of measurements and background produced by various detectors. Methods of elastic tracking (ET) or deformable templates (DT) were developed to satisfy a need in approaches more powerful than conventional tracking methods. It is shown here that both ET and DT methods can be considered as particular cases of a generalized robust approach. A robust method is proposed as a dynamically reweighted iterative least square procedure with a special weight function. Maximum likelihood method is used to optimize this weight function for the given level of contamination. Special study of the joint robust estimate of location and scale parameters provides an effective algorithm to avoid spurious local minima of least square functional. Several examples of effective RM applications are given.

ISO 690:

Ososkov, G. 2003. Elastic arm methods of data analysis as a robust approach. In Tatra Mountains Mathematical Publications, vol. 26, no.2, pp. 291-306. 1210-3195.

APA:

Ososkov, G. (2003). Elastic arm methods of data analysis as a robust approach. Tatra Mountains Mathematical Publications, 26(2), 291-306. 1210-3195.