Tag
This paper introduces NM-PPG, a non-myopic active feature acquisition method using pathwise policy gradients to optimize sequential feature selection in costly prediction scenarios.