从神经坍缩视角重新思考损失重加权，实现更均衡的分类性能。
Rethinking Loss Reweighting for Imbalance Learning as an Inverse Problem: A Neural Collapse Point of View
- 将损失重加权视为逆问题，以等损失为目标动态推导权重。
- 在多个数据集上显著降低损失不均衡系数，优于主流基线方法。
- 适合长尾分布分类任务，尤其关注模型收敛几何性质的研究者。
损失重加权是长尾分类中常用的方法，但现有策略多依赖启发式设计，缺乏明确目标。受神经坍缩（NC）理论启发，理想单纯形等角紧框架（ETF）终端几何表明，各类别平均损失相等是一个合理的重加权目标。基于此等损失目标，我们将损失重加权建模为逆问题，提出一种逆视角重加权策略，动态推断类别权重以逼近该理想状态。实验表明，NC指标显示本方法能有效降低损失不均衡系数，并更接近神经坍缩几何结构，在不同数据集上持续超越强基线模型。
原文摘要 · Abstract (English)
Loss reweighting is a widely used strategy for long-tailed classification, but existing reweighting strategies often rely on heuristics and rarely define a well-specified target. Inspired by Neural Collapse (NC), the ideal simplex Equiangular Tight Frame (ETF) terminal geometry suggests equal per-class average loss as a reasonable target for reweighting. Based on the ideal equal loss objective, we consider loss reweighting as an inverse problem and propose an inverse-view reweighting strategy that infers class weights dynamically to match this ideal objective. Empirically, NC metrics suggest our method can effectively reduce the loss imbalance coefficient and closer alignment with NC geometry while consistently outperforming strong long-tailed baselines on different datasets.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。