English

PriorBoost: An Adaptive Algorithm for Learning from Aggregate Responses

Machine Learning 2024-02-08 v1 Data Structures and Algorithms

Abstract

This work studies algorithms for learning from aggregate responses. We focus on the construction of aggregation sets (called bags in the literature) for event-level loss functions. We prove for linear regression and generalized linear models (GLMs) that the optimal bagging problem reduces to one-dimensional size-constrained kk-means clustering. Further, we theoretically quantify the advantage of using curated bags over random bags. We then propose the PriorBoost algorithm, which adaptively forms bags of samples that are increasingly homogeneous with respect to (unobserved) individual responses to improve model quality. We study label differential privacy for aggregate learning, and we also provide extensive experiments showing that PriorBoost regularly achieves optimal model quality for event-level predictions, in stark contrast to non-adaptive algorithms.

Keywords

Cite

@article{arxiv.2402.04987,
  title  = {PriorBoost: An Adaptive Algorithm for Learning from Aggregate Responses},
  author = {Adel Javanmard and Matthew Fahrbach and Vahab Mirrokni},
  journal= {arXiv preprint arXiv:2402.04987},
  year   = {2024}
}

Comments

29 pages, 4 figures