Semi-supervised learning on data streams with concept drift has attracted considerable attention in recent decades. Many existing algorithms have achieved promising results on low-dimensional data streams by leveraging unlabeled data and adapting to nonstationary distributions. However, real-world data streams often exhibit complex entanglement and high dimensionality, which are usually overlooked by current approaches, resulting in fragile classification performance. This highlights the need for effective representation learning on evolving data streams to enhance the reliability of model prediction. To this end, we propose a novel algorithm for online semi-supervised learning on high-dimensional data streams by learning from contrastive evolving micro-clusters (MCs), named CEMC. Unlike existing methods, CEMC explicitly mitigates feature entanglement through contrastive MC representation learning, with model initialization guided by contrastive objectives and representation updates triggered by potential drift. To ensure reliable semi-supervised learning, we further model the reliability of contrastive MCs to support online classification and enable rapid adaptation to concept drift. By maintaining contrastive MCs online, CEMC preserves and adapts to evolving concepts within a limited memory budget, while sustaining a discriminative representation space. Empirical results on fourteen real-world benchmark datasets demonstrate the effectiveness of CEMC compared to six state-of-the-art algorithms.

Mining feature evolvable streams has gained increasing attention in recent years. However, most existing approaches are designed for stationary data streams (i.e., data streams without concept drifts) and often work with high time complexity due to the time-consuming optimization procedure. The two deficiencies thus largely limit its applications to real-world data stream scenarios. In this paper, we consider a more difficult but practical streaming setting: a data stream with both concept drifts and evolvable features. To this end, we propose a general framework for mining concept-drifting data streams with evolvable features, called FEMC, based on an efficient Feature Evolvable streaming learning and dynamic Micro-Clusters maintenance. Specifically, we derive a closed-form solution to preserve the information in vanished features by learning a weight vector on survival features. The evolving concepts, are further learnt by dynamically maintaining a set of micro-clusters with varying feature space on-the-fly. Empirical results on real-world data sets have demonstrated the benefits of the proposed framework on both clustering and classification tasks by comparing with state-of-the-art algorithms.