Abstract

Gaussian process regression (GPR) is a fundamental model used in machine learning (ML). Due to its accurate prediction with uncertainty and versatility in handling various data structures via kernels, GPR has been successfully used in various applications. However, in GPR, how the features of an input contribute to its prediction cannot be interpreted. Here, we propose GPR with local explanation, which reveals the feature contributions to the prediction of each sample while maintaining the predictive performance of GPR. In the proposed model, both the prediction and explanation for each sample are performed using an easy-to-interpret locally linear model. The weight vector of the locally linear model is assumed to be generated from multivariate Gaussian process priors. The hyperparameters of the proposed models are estimated by maximizing the marginal likelihood. For a new test sample, the proposed model can predict the values of its target variable and weight vector, as well as their uncertainties, in a closed form. Experimental results on various benchmark datasets verify that the proposed model can achieve predictive performance comparable to those of GPR and superior to that of existing interpretable models and can achieve higher interpretability than them, both quantitatively and qualitatively.

Highlights

  • G AUSSIAN processes (GPs) have been well studied for constructing probabilistic models as priors of nonlinear functions in the machine learning (ML) community

  • To overcome the aforementioned limitations, we propose a novel framework for GP-based regression models, Gaussian process regression with local explanation, called GPX, which reveals the feature contributions to the prediction for each sample while maintaining the predictive performance of GPR

  • We focus on generating predictions with explanations using locally linear models

Read more

Summary

Introduction

G AUSSIAN processes (GPs) have been well studied for constructing probabilistic models as priors of nonlinear functions in the machine learning (ML) community. They have demonstrated great success in various problem settings, such as regression [1], [2], classification [1], [3], time-series forecasting [4], and black-box optimization [5]. GPR is defined on an infinite-dimensional feature space via kernel functions. It requires the values of the Manuscript received November 6, 2020; revised June 22, 2021 and November 5, 2021; accepted November 22, 2021.

Objectives
Methods
Findings
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call