Um algoritmo de amostragem para redes de difusão adaptativas multitarefas
Daniel Gilio Tiglea, Renato Candido, Magno T. M. Silva. Um algoritmo de amostragem para redes de difusão adaptativas multitarefas. Simpósio Brasileiro de Telecomunicações – SBrT’2023, 2023, São José dos Campos, Anais do SBrT 2023, 2023, pp. 1–5. (in portuguese)
In this paper, a sampling algorithm for clustered multitask adaptive diffusion networks is proposed. The proposed algorithm seeks to keep the nodes sampled when the squared error in their clusters is high and stops sampling them otherwise. Simulations show that the proposed solution achieves a computational cost similar to that of a single-task algorithm, but with better performance and a convergence rate comparable to that of a multitask algorithm with all nodes sampled. Furthermore, the proposed algorithm prevents adverse effects on a specific cluster from affecting the sampling across the entire network.
Adaptive diffusion networks, multitask learning, sampling, distributed signal processing, impulsive noise.