Decoupling Representation and Classifier for Long-Tailed Recognition 图像领域长尾分布分类问题方法

文章目录

• • • 往期文章链接目录
    • Introduction
    • Recent Directions
    • Sampling Strategies
    • Methods of Learning Classifiers
    • • Classifier Re-training (cRT)
      • Nearest Class Mean classifier (NCM)
      • $\tau$-normalized classifier $(\tau$-normalized)
    • Experiments
    • • Datasets
      • Evaluation Protocol
      • Results
      • • Sampling matters when training jointly
        • Instance-balanced sampling generalize well
        • Decoupled Learning strategy helps
        • Weight Norm Visualization
      • Compare to SOTA
    • Contributions
    • 往期文章链接目录

往期文章链接目录

Introduction

When learning with long-tailed data, a common challenge is that instance-rich (or head) classes dominate the training procedure. The learned classification model tends to perform better on these classes, while performance is significantly worse for instance-scarce (or tail) classes (under-fitting).

The general scheme for long-tailed recognition is: classifiers are either learned jointly with the representations end-to-end, or via a two-stage approach where the classifier and the representation are jointly fine-tuned with variants of class-balanced sampling as a second stage.

In our work, we argue for decoupling representation and classification. We demonstrate that in a long-tailed scenario, this separation allows straightforward approaches to achieve high recognition performance, without the need for designing sampling strategies, balance-aware losses or adding memory modules.

Recent Directions

Recent studies’ directions on solving long-tailed recognition problem:

• Data distribution re-balancing. Re-sample the dataset to achieve a more balanced data distribution. These methods include over-sampling, down-sampling and class-balanced sampling.
• Class-balanced Losses. Assign different losses to different training samples for each class.
• Transfer learning from head- to tail classes. Transfer features learned from head classes with abundant training instances to under-represented tail classes. However it is usually a non-trivial task to design specific modules for feature transfer.

Sampling Strategies

For most sampling strategies presented below, the probability p j p_j pj​ of sampling a data point from class $j$ is given by: p j = n j q ∑ i = 1 C n i q p_{j}=\frac{n_{j}^{q}}{\sum_{i=1}^{C} n_{i}^{q}} pj​=∑i=1C​niq​njq​​ where $q\in [0,1]$, n j n_j nj​ denote the number of training sample for class $j$ and

本文来自互联网用户投稿，文章观点仅代表作者本人，不代表本站立场，不承担相关法律责任。如若转载，请注明出处。 如若内容造成侵权/违法违规/事实不符，请点击【内容举报】进行投诉反馈！