Why Not Hyperparameter-Friendly Optimisation? A Monotonic Adaptive Norm Rescaling Approach For Long-Tailed Recognition 文章

ArXiv CS.CV2026-06-02NEWSen作者: Shuo Zhang, Chenqi Li, Tingting Zhu
查看原文 →
关系图谱

Why Not Hyperparameter-Friendly Optimisation? A Monotonic Adaptive Norm Rescaling Approach For Long-Tailed Recognition · 相关技术

相关技术

parameter regularizationclassifier retrainingtwo-stage decoupling paradigmdeep learningSelf-Adaptive Monotonic Normalizationadaptive norm rescalingLong-tailed recognition