.png){kind=icon}
.png){kind=icon}
Return
Multitask diffusion adaptation over hyper-networks
.png){kind=icon}
PRE
AI
DOI：10.1016/j.dsp.2024.104920.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Current adaptive networks have several challenges in processing various data types, including medical signals, audio signals, telecommunication signals, etc. Researchers have introduced different methods to solve these challenges. However, these methods still show limitations in the face of certain data types. This paper aims to suggest a new framework using an adaptive filtering approach in a novel concept named hyper-network, in which, a set of hyper-nodes interact with each other, each of which can be a unique adaptive network. The proposed adaptive filtering framework can be based on the least means square (LMS) networks and lead to increased accuracy compared to existing techniques. We performed a theoretical analysis on the convergence of the mean and mean square for the proposed framework. Furthermore, we performed experiments to compare it with multiple other approaches suggested in the field, aiming to assess their effectiveness. The obtained results provide strong evidence for the effectiveness of our suggested framework and highlight its potential.
Keywords:
Adaptive filtering
Signal processing
Least mean squares
Adaptive network
Multichannel data