From Neural Collapse to Label-Limited Evolving Streams: Geometry-Constrained Learning Under Dynamic Class Imbalance

Hongliang Wang; Hongyuan Liu; Qirui Hao; Yun Sing Koh; Qinli Yang; Junming Shao

IJCAI 20260 citations

From Neural Collapse to Label-Limited Evolving Streams: Geometry-Constrained Learning Under Dynamic Class Imbalance

Hongliang Wang, Hongyuan Liu, Qirui Hao, Yun Sing Koh, Qinli Yang, Junming Shao

Abstract

Learning from label-limited streams presents significant challenges, particularly when coupled with concept drift and dynamic class imbalance. Existing works often struggle to maintain a discriminative feature space under these constraints, biasing decision boundaries toward majority classes or outdated concepts. To address this, we propose a novel framework named Neural collapse Inspired Label-limited Evolving stream learning (NILE). Instead of utilizing learnable classifiers, NILE exploits Neural Collapse (NC) geometry to explicitly construct a Simplex Equiangular Tight Frame (ETF) as a fixed classifier, ensuring maximal inter-class separability to guide feature discriminability. To maintain this discrimination with limited labels, NILE employs hybrid active learning to prioritize uncertain and minority samples, and an NC-adapted semi-supervised mechanism to enhance representation learning. NILE further updates the classifier by dynamically adjusting the ETF structure, enabling continual adaptation to drifting concepts. Extensive experiments show that NILE can effectively guide features toward the NC state, significantly outperforming state-of-the-art baselines in nonstationary environments.

Data Mining: Mining data streamsData Mining: Class imbalance and unequal costMachine Learning: Semi-supervised learningMachine Learning: Active learning

BibTeX

@inproceedings{ijcai2026_fromneuralcollap,
  title = {From Neural Collapse to Label-Limited Evolving Streams: Geometry-Constrained Learning Under Dynamic Class Imbalance},
  author = {Hongliang Wang and Hongyuan Liu and Qirui Hao and Yun Sing Koh and Qinli Yang and Junming Shao},
  booktitle = {IJCAI 2026},
  year = {2026}
}