Long-Tail Challenge

In the field of artificial intelligence, the Long-Tail Challenge usually refers to a class of problems encountered in machine learning and deep learning, especially when dealing with visual recognition tasks. The Long-Tail Challenge focuses on the class imbalance problem, that is, in the data set, the minority class (head class) has a large number of samples, while the majority class (tail class) has only a few samples. This situation will cause the model to tend to learn the features of high-frequency classes during training and ignore low-frequency classes, thus affecting the performance of the model on the overall data set, especially on rare classes.

In academic research, there are more and more papers on the long-tail challenge. For example, Yan Shuicheng and Feng Jiashi’s team conducted research on deep long-tail learning at the National University of Singapore and SEA AI Lab, and published a related review paper “Deep Long-Tailed Learning: A Survey", systematically expounded deep long-tail learning and its methods and applications, and proposed a new evaluation indicator to verify the ability of existing long-tail learning methods to solve the problem of class imbalance.