Federated Gaussian Process Learning via Pseudo-Representations for Large-Scale Multi-Robot Systems Conference: Proc. of the 25th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2026); May 25 – 29, 2026; Paphos, Cyprus

Sanket A. Salunkhe, George P. Kontoudis · 2026

Read Federated Gaussian Process Learning via Pseudo-Representations for Large-Scale Multi-Robot Systems Conference: Proc. of the 25th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2026); May 25 – 29, 2026; Paphos, Cyprus by Sanket A. Salunkhe, George P. Kontoudis (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-12243.txt.

Description

Robot teams need shared maps of complex places, but Gaussian process models grow cubic in cost and strain links between robots. Sanket A. Salunkhe led this work with one co-author, George P. Kontoudis, presenting pxpGP, a distributed Gaussian process frame for large central and decentralised robot networks. Each robot builds a compact local pseudo-data set through sparse variational inference.

A global consensus method then joins those local pieces, using scaled proximal steps, adaptive settings, and warm starts. Tests on synthetic and real data show pxpGP and its decentralised twin beating existing distributed methods on hyperparameter estimates and prediction accuracy, with the edge growing in large networks. The result is a scalable way for robot teams to model a shared world.

Catalog record

CREATORSanket A. Salunkhe, George P. Kontoudis
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE44 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-12243

Browse the full library · About the Tower of Babel