AI Paper Atlas

Research map / Learning theory and algorithms / Optimal transport

Re-evaluating Word Mover’s Distance

Ryoma Sato, Makoto Yamada, Hisashi Kashima

ICML 2022 · Learning theory and algorithms › Optimal transport · approach: Neural optimal transport · 11 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
L1-normalize BOW vectors before kNN classificationBOW · All eight datasetsrelative kNN classification error: 3.931 -> 1.000Improved
Use optimal transport with word-embedding geometry after L1 normalizationWMD versus BOW (L1/L1) · All eight datasetsrelative kNN classification error: 1.000 -> 0.917Improved
Use WMD instead of the strongest L1/L1 classical baselineWMD · bbcsport, twitter, recipe, ohsumed, classic, reuters, amazon, and 20newskNN classification errorMixed
Use TF-IDF weighting for WMD marginalsWMD · Eight document-classification datasetskNN classification error: approximately a six percent improvement from the normalized TF-IDFImproved
Use TF-IDF weighting for WMD marginalsWMD · Clean datasetskNN classification error: two percent improvementImproved
Remove duplicate documents from the evaluation dataBOW, TF-IDF, and WMD · Eight document-classification datasetskNN classification errorNo clear effect
Include words outside the word2vec vocabulary in BOW and TF-IDFBOW and TF-IDF · All-word datasetskNN classification errorNo clear effect
Use 5-dimensional instead of 300-dimensional embeddings (ablation)WMD · Eight document-classification datasetskNN classification errorHurt
Vary embedding normalization and distance metric in WMD's cost function (ablation)WMD · Eight document-classification datasetskNN classification errorNo clear effect
Use exponential distance weighting in weighted kNN instead of standard kNN votingBOW, TF-IDF, WMD, and WMD-TF-IDF · Eight document-classification datasetsclassification errorImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy