Aggressive Sampling for Multi-class to Binary Reduction with Applications to Text Classification
This addresses the problem of efficient and scalable multi-class classification for applications like text classification, but it is incremental as it builds on existing reduction strategies with a sampling enhancement.
The paper tackles multi-class classification with a large number of classes by proposing a double sampling strategy combined with a multi-class to binary reduction, which transforms the problem into binary classification over pairs of examples. Experiments on datasets with 10,000 to 100,000 classes show improvements in training time, prediction time, memory consumption, and predictive performance compared to state-of-the-art methods.
We address the problem of multi-class classification in the case where the number of classes is very large. We propose a double sampling strategy on top of a multi-class to binary reduction strategy, which transforms the original multi-class problem into a binary classification problem over pairs of examples. The aim of the sampling strategy is to overcome the curse of long-tailed class distributions exhibited in majority of large-scale multi-class classification problems and to reduce the number of pairs of examples in the expanded data. We show that this strategy does not alter the consistency of the empirical risk minimization principle defined over the double sample reduction. Experiments are carried out on DMOZ and Wikipedia collections with 10,000 to 100,000 classes where we show the efficiency of the proposed approach in terms of training and prediction time, memory consumption, and predictive performance with respect to state-of-the-art approaches.