Gaussian process classification for variable fidelity data

Abstract

In this paper we address a classification problem where two sources of labels with different levels of fidelity are available. Our approach is to combine data from both sources by applying a co-kriging schema on latent functions, which allows the model to account item-dependent labeling discrepancy. We provide an extension of Laplace inference for Gaussian process classification, that takes into account multi-fidelity data. We evaluate the proposed method on real and synthetic datasets and show that it is more resistant to different levels of discrepancy between sources than other approaches for data fusion. Our method can provide accuracy/cost trade-off for a number of practical tasks such as crowd-sourced data annotation and feasibility regions construction in engineering design.

Learning Multi-Task Gaussian Process Over Heterogeneous Input Domains

“…Thereafter, similar to (8), by reformulating the KL divergence KL[q(f , u, X)||p(f , u, X|y)], we arrive at the ELBO expressed as…”

“…Different from the conventional single-task GP [3], the MTGP particularly exploits and represents the correlations among related tasks in order to achieve knowledge sharing and transfer, which therefore improves the quality of prediction and alleviates the demand of large-scale training data. Consequently, the MTGPs have gained widespread application in diverse domains, for example, time series forecasting [4], multi-task optimization [5]- [7] and multi-fidelity classification [8].…”