Online Gaussian Process State-Space Model: Learning and Planning for Partially Observable Dynamical Systems

TL;DR

This paper introduces an online learning approach for Gaussian process state-space models that efficiently updates with sequential data, enabling adaptive modeling and planning in partially observable dynamical systems.

Contribution

It presents a novel online variational inference method for GP-SSMs that reduces computational costs and supports adaptation, extending their application to reinforcement learning.

Findings

Effective online learning with reduced computational resources.

Supports adaptation to system changes and real environments.

Demonstrated applicability in reinforcement learning scenarios.

Abstract

This paper proposes an online learning method of Gaussian process state-space model (GP-SSM). GP-SSM is a probabilistic representation learning scheme that represents unknown state transition and/or measurement models as Gaussian processes (GPs). While the majority of prior literature on learning of GP-SSM are focused on processing a given set of time series data, data may arrive and accumulate sequentially over time in most dynamical systems. Storing all such sequential data and updating the model over entire data incur large amount of computational resources in space and time. To overcome this difficulty, we propose a practical method, termed \textit{onlineGPSSM}, that incorporates stochastic variational inference (VI) and online VI with novel formulation. The proposed method mitigates the computational complexity without catastrophic forgetting and also support adaptation to changes in a system and/or a real environments. Furthermore, we present application of onlineGPSSM into the reinforcement learning (RL) of partially observable dynamical systems by integrating onlineGPSSM with Bayesian filtering and trajectory optimization algorithms. Numerical examples are presented to demonstrate applicability of the proposed method.