Memory-based Jitter: Improving Visual Recognition on Long-tailed Data with Diversity In Memory
This addresses the challenge of improving visual recognition for long-tailed data, which is common in real-world applications, by enhancing diversity in tail classes, though it is incremental as it builds on existing memory-based and augmentation techniques.
The paper tackles the problem of limited within-class diversity in tail classes for long-tailed visual recognition by introducing Memory-based Jitter (MBJ), which collects weight and feature jitters from historical model editions to augment tail classes, achieving performance on par with state-of-the-art on classification and metric learning benchmarks.
This paper considers deep visual recognition on long-tailed data. To be general, we consider two applied scenarios, \ie, deep classification and deep metric learning. Under the long-tailed data distribution, the majority classes (\ie, tail classes) only occupy relatively few samples and are prone to lack of within-class diversity. A radical solution is to augment the tail classes with higher diversity. To this end, we introduce a simple and reliable method named Memory-based Jitter (MBJ). We observe that during training, the deep model constantly changes its parameters after every iteration, yielding the phenomenon of \emph{weight jitters}. Consequentially, given a same image as the input, two historical editions of the model generate two different features in the deeply-embedded space, resulting in \emph{feature jitters}. Using a memory bank, we collect these (model or feature) jitters across multiple training iterations and get the so-called Memory-based Jitter. The accumulated jitters enhance the within-class diversity for the tail classes and consequentially improves long-tailed visual recognition. With slight modifications, MBJ is applicable for two fundamental visual recognition tasks, \emph{i.e.}, deep image classification and deep metric learning (on long-tailed data). Extensive experiments on five long-tailed classification benchmarks and two deep metric learning benchmarks demonstrate significant improvement. Moreover, the achieved performance are on par with the state of the art on both tasks.