Rethinking Loss Reweighting for Imbalance Learning as an Inverse Problem: A Neural Collapse Point of View
For practitioners dealing with imbalanced datasets, this work provides a principled reweighting approach grounded in Neural Collapse theory, offering improved performance over heuristic methods.
The paper addresses long-tailed classification by rethinking loss reweighting through the lens of Neural Collapse, proposing an inverse-view reweighting strategy that dynamically infers class weights to achieve equal per-class average loss. The method consistently outperforms strong baselines across multiple datasets.
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.