AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Spherical Text Embedding

Yu Meng, Jiaxin Huang, Guangyuan Wang, Chao Zhang, Honglei Zhuang, Lance Kaplan, Jiawei Han

NeurIPS 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 51 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
Train embeddings directly in spherical space rather than Euclidean spaceJoSE word embeddings · Word similarity: WordSim353, MEN, SimLex999Spearman rank correlation: WordSim353: 0.711 -> 0.739; MEN: 0.726 -> 0.748; SimLex999: 0.321 -> 0.339Improved
Use JoSE paragraph embeddings instead of the strongest listed baselineDocument clustering with K-Means · 20 NewsgroupMI: 1.856 ± 0.020 -> 1.975 ± 0.026Improved
Use JoSE paragraph embeddings instead of the strongest listed baselineDocument clustering with SK-Means · 20 NewsgroupNMI: 0.630 ± 0.007 -> 0.664 ± 0.010Improved
Use JoSE paragraph embeddings instead of the strongest listed baselineDocument clustering with SK-Means · 20 NewsgroupARI: 0.494 ± 0.012 -> 0.568 ± 0.020Improved
Use JoSE paragraph embeddings instead of the strongest listed baselineDocument clustering with SK-Means · 20 NewsgroupPurity: 0.648 ± 0.017 -> 0.721 ± 0.029Improved
Use spherical K-Means instead of K-MeansJoSE paragraph embeddings · 20 NewsgroupNMI: 0.663 ± 0.008 -> 0.664 ± 0.010No clear effect
Use JoSE paragraph embeddings instead of Avg. W2V20 Newsgroup k-NN classification · 20 NewsgroupMacro-F1: 0.630 -> 0.703Improved
Use JoSE paragraph embeddings instead of Avg. W2V20 Newsgroup k-NN classification · 20 NewsgroupMicro-F1: 0.631 -> 0.707Improved
Use JoSE paragraph embeddings instead of Avg. W2VMovie Review k-NN classification · Movie ReviewMacro-F1: 0.712 -> 0.764Improved
Use JoSE paragraph embeddings instead of Avg. W2VMovie Review k-NN classification · Movie ReviewMicro-F1: 0.713 -> 0.765Improved
Train with JoSE rather than Word2VecWord embedding training · Wikipedia dumpTraining time per iteration: 0.81 hrs -> 0.73 hrsImproved

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