LGAINov 25, 2025

Space Alignment Matters: The Missing Piece for Inducing Neural Collapse in Long-Tailed Learning

arXiv:2512.07844v11 citations
Originality Incremental advance
AI Analysis

This addresses the generalization performance issue in long-tailed learning for computer vision applications, representing an incremental improvement by enhancing existing methods.

The paper tackles the problem of neural collapse not occurring in long-tailed learning due to misalignment between feature and classifier weight spaces, proposing three plug-and-play alignment strategies that achieve state-of-the-art performance on datasets like CIFAR-10-LT, CIFAR-100-LT, and ImageNet-LT.

Recent studies on Neural Collapse (NC) reveal that, under class-balanced conditions, the class feature means and classifier weights spontaneously align into a simplex equiangular tight frame (ETF). In long-tailed regimes, however, severe sample imbalance tends to prevent the emergence of the NC phenomenon, resulting in poor generalization performance. Current efforts predominantly seek to recover the ETF geometry by imposing constraints on features or classifier weights, yet overlook a critical problem: There is a pronounced misalignment between the feature and the classifier weight spaces. In this paper, we theoretically quantify the harm of such misalignment through an optimal error exponent analysis. Built on this insight, we propose three explicit alignment strategies that plug-and-play into existing long-tail methods without architectural change. Extensive experiments on the CIFAR-10-LT, CIFAR-100-LT, and ImageNet-LT datasets consistently boost examined baselines and achieve the state-of-the-art performances.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes