PDL Abstract

DriftSurf: A Risk-competitive Learning Algorithm under Concept Drift

arXiv:2003.06508 [cs.LG], August, 2020.

Ashraf Tahmasbi1 Ellango Jothimurugesan2, Srikanta Tirthapura1,3, Phillip B. Gibbons2

1Department of Electrical and Computer Engineering, Iowa State University
2Computer Science Department, Carnegie Mellon University
3Apple

http://www.pdl.cmu.edu/

When learning from streaming data, a change in the data distribution, also known as concept drift, can render a previously-learned model inaccurate and require training a new model. We present an adaptive learning algorithm that extends previous drift-detection-based methods by incorporating drift detection into a broader stable-state/reactivestate process. The advantage of our approach is that we can use aggressive drift detection in the stable state to achieve a high detection rate, but mitigate the false positive rate of standalone drift detection via a reactive state that reacts quickly to true drifts while eliminating most false positives. The algorithm is generic in its base learner and can be applied across a variety of supervised learning problems. Our theoretical analysis shows that the risk of the algorithm is competitive to an algorithm with oracle knowledge of when (abrupt) drifts occur. Experiments on synthetic and real datasets with concept drifts confirm our theoretical analysis.

FULL PAPER: pdf