PrO-GPs：以预测不确定性为目标的 Gaussian Process 新方法
Predictively Oriented Gaussian Process Posteriors
做回归或不确定性量化的人可以看看：标准 GP 一旦 kernel 选错就崩，PrO-GPs 直接优化预测校准，实验显示误设下表现更稳。
论文提出 Predictively Oriented Gaussian Processes（PrO-GPs），把预测不确定性作为主要推断目标，替代标准 GP 需要先选 kernel 和观测模型的做法。由于非参数模型的 PrO 后验无法直接计算，作者推导了简化形式和实用的采样方案以高效求解。在合成数据与真实数据实验中，PrO-GPs 在模型误设（misspecification）情况下比标准 GP 产生校准更好的预测分布。
Predictively Oriented Gaussian Process Posteriors
Gaussian Processes (GPs) are a powerful tool for modelling and quantifying uncertainty in functional relationships. However, they require practitioners to make a number of design decisions, such as the choice of the kernel and the observation model. Suboptimal choices can produce misspecified models that do not capture the underlying data generating process. We introduce Predictively Oriented Gaussian Processes (PrO-GPs), which treat predictive uncertainty as the primary inferential target and provide a robust alternative to standard GPs. Although direct computation of a PrO posterior for nonparametric models is intractable, we derive a reduced formulation and practical sampling scheme for efficient computation. Through synthetic and real data experiments, we show that PrO-GPs produce better calibrated predictive distributions under model misspecification compared to standard GP approaches.