Nonnegativity-Enforced Gaussian Process Regression
This addresses a domain-specific issue for applications requiring physically bounded models, but appears incremental as it modifies an existing framework.
The paper tackled the problem of standard Gaussian Process regression producing unbounded models that can take infeasible values for processes with physical constraints, by proposing an approach to enforce these constraints probabilistically, which also reduces variance in the resulting model.
Gaussian Process (GP) regression is a flexible non-parametric approach to approximate complex models. In many cases, these models correspond to processes with bounded physical properties. Standard GP regression typically results in a proxy model which is unbounded for all temporal or spacial points, and thus leaves the possibility of taking on infeasible values. We propose an approach to enforce the physical constraints in a probabilistic way under the GP regression framework. In addition, this new approach reduces the variance in the resulting GP model.