Communication-aware Distributed Gaussian Process Regression Algorithms for Real-time Machine Learning

Zhenyuan Yuan, Minghui Zhu

Abstract: We propose a communication-aware Gaussian process regression algorithm that allows a network of robots to collaboratively learn about a common latent function in real time using streaming data. We quantify the improvement that inter-robot communication brings on the transient performance of the learning algorithm. Simulations are performed to validate the proposed algorithm.