On the Dynamics of Multiagent Nonlinear Filtering and Learning

Sayed Talebi; Danilo Mandic

doi:10.1109/MLSP58920.2024.10734830

On the Dynamics of Multiagent Nonlinear Filtering and Learning

Sayed Talebi, Danilo Mandic

School of Arts, Humanities, and Social Sciences

Research output: Contribution to journal › Article › peer-review

Abstract

Multiagent systems aim to accomplish highly complex learning tasks through decentralised consensus seeking dynamics. In the past decade, use of the multiagent framework either in the form of federated learning or distributed estimation has garnered a great deal of attention in the signal processing and computational intelligence societies. This article examines behaviour of multiagent networked systems that use nonlinear filtering/learning frameworks. This article considers the most general learning framework that is reliant on multiple input streams. Operations of this centralised framework is decomposed into operations corresponding to the federated and distributed frameworks. This decomposition process is used to formulate the changes in learning dynamics as the framework shifts from centralised to distributed. Importantly, the formulated dynamics indicate convergence conditions of distributed learning techniques are much more relaxed than formerly considered.

Original language	English
Journal	IEEE International Workshop on Machine Learning for Signal Processing
DOIs	https://doi.org/10.1109/MLSP58920.2024.10734830
Publication status	Published - 4 Nov 2024

Access to Document

10.1109/MLSP58920.2024.10734830

Cite this

@article{25954ae8f2094b8eaa61fa1bde78b22e,

title = "On the Dynamics of Multiagent Nonlinear Filtering and Learning",

abstract = "Multiagent systems aim to accomplish highly complex learning tasks through decentralised consensus seeking dynamics. In the past decade, use of the multiagent framework either in the form of federated learning or distributed estimation has garnered a great deal of attention in the signal processing and computational intelligence societies. This article examines behaviour of multiagent networked systems that use nonlinear filtering/learning frameworks. This article considers the most general learning framework that is reliant on multiple input streams. Operations of this centralised framework is decomposed into operations corresponding to the federated and distributed frameworks. This decomposition process is used to formulate the changes in learning dynamics as the framework shifts from centralised to distributed. Importantly, the formulated dynamics indicate convergence conditions of distributed learning techniques are much more relaxed than formerly considered.",

author = "Sayed Talebi and Danilo Mandic",

year = "2024",

month = nov,

day = "4",

doi = "10.1109/MLSP58920.2024.10734830",

language = "English",

journal = "IEEE International Workshop on Machine Learning for Signal Processing",

}

TY - JOUR

T1 - On the Dynamics of Multiagent Nonlinear Filtering and Learning

AU - Talebi, Sayed

AU - Mandic, Danilo

PY - 2024/11/4

Y1 - 2024/11/4

N2 - Multiagent systems aim to accomplish highly complex learning tasks through decentralised consensus seeking dynamics. In the past decade, use of the multiagent framework either in the form of federated learning or distributed estimation has garnered a great deal of attention in the signal processing and computational intelligence societies. This article examines behaviour of multiagent networked systems that use nonlinear filtering/learning frameworks. This article considers the most general learning framework that is reliant on multiple input streams. Operations of this centralised framework is decomposed into operations corresponding to the federated and distributed frameworks. This decomposition process is used to formulate the changes in learning dynamics as the framework shifts from centralised to distributed. Importantly, the formulated dynamics indicate convergence conditions of distributed learning techniques are much more relaxed than formerly considered.

AB - Multiagent systems aim to accomplish highly complex learning tasks through decentralised consensus seeking dynamics. In the past decade, use of the multiagent framework either in the form of federated learning or distributed estimation has garnered a great deal of attention in the signal processing and computational intelligence societies. This article examines behaviour of multiagent networked systems that use nonlinear filtering/learning frameworks. This article considers the most general learning framework that is reliant on multiple input streams. Operations of this centralised framework is decomposed into operations corresponding to the federated and distributed frameworks. This decomposition process is used to formulate the changes in learning dynamics as the framework shifts from centralised to distributed. Importantly, the formulated dynamics indicate convergence conditions of distributed learning techniques are much more relaxed than formerly considered.

U2 - 10.1109/MLSP58920.2024.10734830

DO - 10.1109/MLSP58920.2024.10734830

M3 - Article

JO - IEEE International Workshop on Machine Learning for Signal Processing

JF - IEEE International Workshop on Machine Learning for Signal Processing

ER -