Properties of the Weighted and Robust Implicitly Weighted Correlation Coefficients

0577080 - ÚI 2024 RIV CH eng C - Conference Paper (international conference)
Kalina, Jan - Vidnerová, Petra
Properties of the Weighted and Robust Implicitly Weighted Correlation Coefficients.
Artificial Neural Networks and Machine Learning – ICANN 2023. Proceedings, Part IX. Cham: Springer, 2023 - (Iliadis, L.; Papaleonidas, A.; A, P.; Jayne, C.), s. 200-212. Lecture Notes in Computer Science, 14262. ISBN 978-3-031-44200-1. ISSN 0302-9743.
[ICANN 2023: International Conference on Artificial Neural Networks /32./. Heraklion (GR), 26.09.2023-29.09.2023]
R&D Projects: GA ČR(CZ) GA22-02067S
Institutional support: RVO:67985807
Keywords : Correlation coefficient * Outliers * Robustness * Image analysis * Approximate computing
OECD category: Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)
https://dx.doi.org/10.1007/978-3-031-44201-8_17

Pearson product-moment correlation coefficient represents a fundamental measure of similarity between two data vectors. In various applications, it is meaningful to consider its weighted version known as the weighted Pearson correlation coefficient. Its properties are studied in this theoretical paper - these include the robustness to rounding, as it is an important issue in approximate neurocomputing, or specific robustness properties for the context of template matching in image analysis. For a highly robust correlation coefficient inspired by the least weighted estimator, properties are derived and novel hypothesis tests are proposed. This robust measure is recommendable particularly for data contaminated by outliers (not only) in the context of image analysis.
Permanent Link: https://hdl.handle.net/11104/0346342