Gaussian Processes

2 items

RESEARCHarXiv CS.LG·1d ago

Gaussian Process Latent Factor Regression for Low-Data, High-Dimensional Output Problems

This paper proposes Gaussian Process Latent Factor Regression (GPLFR), a model designed for predicting high-dimensional outputs from few training examples. It couples compression and prediction in a single objective to handle high dimensionality. GPLFR is demonstrated by building the first spatially resolved emulator of global climate models for rocky exoplanets.

Exoplanets Gaussian Processes machine learning Climate Models

RESEARCHarXiv CS.LG·20h ago

Boundary Variance Inflation Causes Acquisition Bias in Gaussian Processes

This paper investigates inflated posterior variance near the boundary in Gaussian processes, tracing the root cause to the truncation of the kernel correlation neighborhood. It shows how this geometric distortion creates acquisition bias, affecting selection patterns across different acquisition classes, independent of objective functions.

Bayesian Optimization Gaussian Processes Statistical Models machine learning