English

Indirect Active Learning

Statistics Theory 2023-01-24 v3 Information Theory Machine Learning math.IT Methodology Machine Learning Statistics Theory

Abstract

Traditional models of active learning assume a learner can directly manipulate or query a covariate XX in order to study its relationship with a response YY. However, if XX is a feature of a complex system, it may be possible only to indirectly influence XX by manipulating a control variable ZZ, a scenario we refer to as Indirect Active Learning. Under a nonparametric model of Indirect Active Learning with a fixed budget, we study minimax convergence rates for estimating the relationship between XX and YY locally at a point, obtaining different rates depending on the complexities and noise levels of the relationships between ZZ and XX and between XX and YY. We also identify minimax rates for passive learning under comparable assumptions. In many cases, our results show that, while there is an asymptotic benefit to active learning, this benefit is fully realized by a simple two-stage learner that runs two passive experiments in sequence.

Keywords

Cite

@article{arxiv.2206.01454,
  title  = {Indirect Active Learning},
  author = {Shashank Singh},
  journal= {arXiv preprint arXiv:2206.01454},
  year   = {2023}
}

Comments

To appear in proceedings of the 26th International Conference on Artificial Intelligence and Statistics (AISTATS)