Speeding up Word Mover's Distance and its variants via properties of distances between embeddings
This work addresses a computational bottleneck for researchers and practitioners using semantic document distances in classification tasks, but it is incremental as it builds on existing variants like RWMD.
The paper tackles the high computational complexity of Word Mover's Distance (WMD) and its variants, which limits their practicality for large datasets, by proposing an approach that speeds up both WMD and Relaxed Word Mover's Distance (RWMD) based on empirical properties of embedding distances. Experiments on 10 datasets show a significant speed-up while maintaining the same error rates.
The Word Mover's Distance (WMD) proposed by Kusner et al. is a distance between documents that takes advantage of semantic relations among words that are captured by their embeddings. This distance proved to be quite effective, obtaining state-of-art error rates for classification tasks, but is also impracticable for large collections/documents due to its computational complexity. For circumventing this problem, variants of WMD have been proposed. Among them, Relaxed Word Mover's Distance (RWMD) is one of the most successful due to its simplicity, effectiveness, and also because of its fast implementations. Relying on assumptions that are supported by empirical properties of the distances between embeddings, we propose an approach to speed up both WMD and RWMD. Experiments over 10 datasets suggest that our approach leads to a significant speed-up in document classification tasks while maintaining the same error rates.