The Application of Variance In Signal’s Denoising Based On ADWA

Attribute Distance Weighted Averaging (ADWA) is a new filtering paradigm whose core mechanism relies on the dynamic adjustment of signal attribute combinations, progressively alleviating the inherent trade-off between signal denoising and feature preservation in traditional filtering algorithms through the continuous introduction of new attributes or replacement of old ones. Within this paradigm, attribute selection plays a critical role in determining filtering outcomes due to the significant differences in the informational characteristics carried by distinct attributes. This study focuses on the "variance" attribute’s role in ADWA-based signal-noise separation, systematically evaluating its contribution to filtering performance and investigating its potential substitution with the " gradient " attribute. The results demonstrate that the ADWA filtering paradigm effectively enhances filtering performance by introducing significant new attributes, achieving superior results in both denoising and feature preservation. This study provides concrete examples for mining new attributes of signals in ADWA paradigm.