Robust, Online, and Adaptive Decentralized Gaussian Processes
This work addresses the problem of scalable and adaptive modeling in decentralized settings for applications like Earth systems, but it is incremental as it builds on an existing decentralized GP framework.
The paper tackled the limitations of Gaussian processes in large-scale, dynamic, and noisy environments by extending a decentralized algorithm with robust filtering and dynamic adaptation, resulting in enhanced stability and accuracy for in-situ modeling in a large-scale Earth system application.
Gaussian processes (GPs) offer a flexible, uncertainty-aware framework for modeling complex signals, but scale cubically with data, assume static targets, and are brittle to outliers, limiting their applicability in large-scale problems with dynamic and noisy environments. Recent work introduced decentralized random Fourier feature Gaussian processes (DRFGP), an online and distributed algorithm that casts GPs in an information-filter form, enabling exact sequential inference and fully distributed computation without reliance on a fusion center. In this paper, we extend DRFGP along two key directions: first, by introducing a robust-filtering update that downweights the impact of atypical observations; and second, by incorporating a dynamic adaptation mechanism that adapts to time-varying functions. The resulting algorithm retains the recursive information-filter structure while enhancing stability and accuracy. We demonstrate its effectiveness on a large-scale Earth system application, underscoring its potential for in-situ modeling.