Jump to Content

Rank4Class: A Ranking Formulation for Multiclass Classification


Multiclass classification (MCC) is a fundamental machine learning problem of classifying each instance into one of a predefined set of classes. Given an instance, an MCC model computes a score for each class, all of which are used to sort the classes. The performance of a model is usually measured by Top-K Accuracy/Error (e.g. K=1 or 5). In this paper, we do not aim to propose new neural network architectures as most recent works do, but to show that it is promising to boost MCC performance with a novel formulation through the lens of ranking. In particular, by viewing MCC as \emph{an instance class ranking problem}, we first argue that ranking metrics, such as Normalized Discounted Cumulative Gain, can be more informative than the existing Top-K metrics. We further demonstrate that the dominant neural MCC recipe can be transformed to a neural ranking pipeline. Based on such generalization, we show that it is intuitive to leverage techniques from the rich information retrieval literature to improve the MCC performance out of the box. Extensive empirical results on both text and image classification tasks with diverse datasets and backbone neural models show the value of our proposed framework.

Research Areas