DT-GOL: Dual-Track Geometric Online Learning in Nonstationary Environment with Label Delay

2026-06-22

AI Summary

This work addresses the challenge of online learning in non-stationary environments where label delay and concept drift jointly hinder timely model adaptation. The authors propose a dual-track geometric online learning framework that treats delayed labels as a semi-supervised task and leverages the real-time topological evolution of feature space as a geometric proxy for concept drift. A dynamic evidence calibration mechanism is introduced to generate uncertainty-aware soft pseudo-labels. By decoupling the dual-track architecture to separately optimize stability and adaptability, the approach effectively mitigates confirmation bias and alleviates the stability-plasticity dilemma. Experimental results demonstrate that the method significantly outperforms existing approaches on both real-world and synthetic datasets, particularly excelling in scenarios where concept drift and label delay co-occur.

Abstract

Online learning is crucial for handling complex data streams in big data applications. Recent research has begun to focus on dynamic scenarios, i.e., non-stationary environments. However, a crucial yet often overlooked aspect is label latency, where new data may not receive labels in time due to the slow and expensive labeling process, thus hindering rapid adaptation to dynamic environments. To resolve this impasse, we propose Dual-Track Geometry Online Learning (DT-GOL), a novel framework that shifts from temporal compensation to spatial reasoning to bridge the supervised latency gap. By modeling the delay challenge as a semi-supervised task, we leverage real-time topological evolution of features as a reliable geometric surrogate for unobservable conceptual changes to achieve proactive supervised adaptation within the delay window. Unlike rigid self-training, we introduce a dynamic evidence calibration mechanism that distills geometric information into soft labels that perceive uncertainty, effectively mitigating the confirmation bias inherent in hard pseudo-labels. Furthermore, to resolve the stability-plasticity dilemma, we design a decoupled dual-track architecture in which a master learner serves as a stable anchor, updated strictly from delayed ground truth, while a transient branch leverages soft geometric knowledge for low-risk forward adaptation. Extensive experiments on real and synthetic datasets demonstrate that DT-GOL significantly outperforms existing state-of-the-art baseline methods, especially in scenarios with concept drift.

Problem

• online learning
• nonstationary environment
• label delay
• concept drift
• semi-supervised learning

Innovation

• geometric online learning
• label delay
• nonstationary environment
• dual-track architecture
• soft pseudo-labels