Generating Synthetic Multi-label Data Streams

Abstract

There are many available methods for generating synthetic data streams. Such methods have been justified by the need to study the efficacy of algorithms on a theoretically infinite stream, and also a lack of real-world data of sufficient size. Although multi-label classification has attracted considerable interest in recent years, most of this work has been carried out in the context of a batch learning environment rather than a data stream. This paper makes an in-depth analysis of multi-label data, and presents a general framework for generating synthetic multi-label data streams.