Abstract
The objective function of numerous well-established Independent Component Analysis (ICA) algorithms calculate based on specific dependence criteria. This study introduces a distinctive dependence criterion based on the cumulative distribution function (CDF) for characterizing the independence between two random variables and some of its properties are examined. Then, we propose a class of ICA algorithms based on the introduced dependence criterion. The performance of the algorithm is systematically compared to some previous similar algorithms. The results indicate that the suggested algorithm have fruitful performance rather than some similar previous known algorithms. Subsequently, the proposed algorithms are applied to real-time series data, serving as an effective pre-processing clustering method.
Recommended Citation
Asadi, Fatemeh; Torabi, Hamzeh; and Nadeb, Hossein
(2025).
(R2119) New Algorithms for Independent Component Analysis Based on a General Class of Dependence Criteria,
Applications and Applied Mathematics: An International Journal (AAM), Vol. 20,
Iss.
1, Article 8.
Available at:
https://digitalcommons.pvamu.edu/aam/vol20/iss1/8