Detail publikace

Evaluation of Background Noise for Significance Level Identification

Originální název

Evaluation of Background Noise for Significance Level Identification

Anglický název

Evaluation of Background Noise for Significance Level Identification

Jazyk

en

Originální abstrakt

The paper deals with identification of significance level for testing the time-frequency transform of the data. The usual procedure of time-frequency significance testing is based on the knowledge of background spectrum. Very often we have expectation of the background noise character (Gaussian noise, Red noise, etc.). Our paper deals with the case when character of the noise is unknown and may not be Gaussian despite our assumptions. Thus, we propose how to identify own critical values for testing time-frequency transform significance with respect to the data character. We compare our knowledge to critical quantile of chi-square distribution.

Anglický abstrakt

The paper deals with identification of significance level for testing the time-frequency transform of the data. The usual procedure of time-frequency significance testing is based on the knowledge of background spectrum. Very often we have expectation of the background noise character (Gaussian noise, Red noise, etc.). Our paper deals with the case when character of the noise is unknown and may not be Gaussian despite our assumptions. Thus, we propose how to identify own critical values for testing time-frequency transform significance with respect to the data character. We compare our knowledge to critical quantile of chi-square distribution.

BibTex


@inproceedings{BUT134601,
  author="Jitka {Poměnková} and Eva {Klejmová} and Tobiáš {Malach}",
  title="Evaluation of Background Noise for Significance
Level Identification",
  annote="The paper deals with identification of significance level for testing the time-frequency transform of the data. The usual procedure of time-frequency significance testing is based on the knowledge of background spectrum. Very often we have expectation of the background noise character (Gaussian noise, Red noise, etc.). Our paper deals with the case when character of the noise is unknown and may not be Gaussian despite our
assumptions. Thus, we propose how to identify own critical values for testing time-frequency transform significance with respect to the data character. We compare our knowledge to critical quantile of chi-square distribution.
",
  booktitle="Proceedings of the 24th International Conference on Systems, Signals and Image Processing",
  chapter="134601",
  howpublished="electronic, physical medium",
  year="2017",
  month="may",
  pages="1--5",
  type="conference paper"
}