Inference-Time Prior Adaptation for Imbalanced Data Streams under Concept Drift

Computer Science > Machine Learning arXiv:2610.33126 (cs) [Submitted on 23 Oct 2026] Title: Inference-Time Prior Adaptation for Imbalanced Data Streams under Concept Drift Authors: Tobias Callies (1), Eirini Ntoutsi (1), Arthur Zimek (2) ((1) University of the Bundeswehr Munich, (2) University of Southern Denmark)

计算机科学 > 机器学习 arXiv:2610.33126 (cs) [提交于 2026 年 10 月 23 日] 标题：概念漂移下不平衡数据流的推理时先验自适应 作者：Tobias Callies (1), Eirini Ntoutsi (1), Arthur Zimek (2) ((1) 慕尼黑联邦国防军大学, (2) 南丹麦大学)

Abstract: Data stream classification under concept drift and class imbalance requires models to adapt continuously while maintaining strong minority-class performance. Many existing approaches rely on active memory management, including instance removal, memory restructuring, resampling, and synthetic data generation.

摘要：在概念漂移和类别不平衡下的数据流分类要求模型在持续适应的同时，保持对少数类别的良好性能。许多现有方法依赖于主动内存管理，包括实例删除、内存重构、重采样和合成数据生成。

We present PAI-KNN, a k-nearest neighbor classifier that adapts via inference-time prior adjustment rather than memory modification. PAI-KNN maintains a simple sliding-window buffer and dynamically adjusts its decision rule using class prior estimates derived from recent observations.

我们提出了 PAI-KNN，这是一种通过推理时先验调整而非内存修改来进行自适应的 k-近邻分类器。PAI-KNN 维护一个简单的滑动窗口缓冲区，并利用从近期观测中得出的类别先验估计来动态调整其决策规则。

Evaluated on a multitude of synthetic data streams covering various imbalance and drift scenarios, PAI-KNN achieves competitive performance with state-of-the-art methods, performing particularly well under scenarios involving dynamic class imbalance and concept drift. These findings highlight the potential of inference-time adaptation in challenging non-stationary environments and suggest that adaptation in memory-based stream classifiers need not rely exclusively on memory modification.

在涵盖各种不平衡和漂移场景的大量合成数据流上进行评估后，PAI-KNN 实现了与最先进方法相当的性能，在涉及动态类别不平衡和概念漂移的场景中表现尤为出色。这些发现突显了推理时自适应在具有挑战性的非平稳环境中的潜力，并表明基于内存的流分类器中的自适应不一定非要依赖于内存修改。