AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Near-Lossless Binarization of Word Embeddings

Julien Tissier, Christophe Gravier, Amaury Habrard

AAAI 2019 · Architectures and efficiency › RNNs and normalization · approach: Learned embeddings for retrieval

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
binarize GloVe vectors to 256 bitsGloVe top-10 search · Top 10 queryexecution time: 98.08 -> 3.25Improved
store binary rather than real-valued GloVe vectorsGloVe top-10 query including loading · Loading + Top 10 queryexecution time: 23500 -> 310Improved
binarize fasttext vectors to 512 bitsfasttext · WS353Spearman correlation: 69.7 -> 70.3Improved
binarize GloVe vectors to 256 bitsGloVe · SimVerbSpearman correlation: 22.7 -> 22.9Improved
binarize fasttext vectors to 512 bitsfasttext · DBpediaaccuracy: 95.0 -> 97.3Improved
binarize fasttext vectors to 512 bitsfasttext · Amazon Fullaccuracy: 49.0 -> 49.8Improved
use the proposed encoder instead of naive binarizationdict2vec · Semantic similarity average, 64-bitFisher's average semantic score: 44.7 -> 46.5Improved
use the proposed encoder instead of naive binarizationdict2vec · Semantic similarity average, 128-bitFisher's average semantic score: 49.6 -> 54.1Improved
increase binary code size from 64 to 256 or 512 bits (ablation)binarized embeddings · semantic similarity and classification taskstask performanceImproved
remove the correlation regularization term (ablation)autoencoder binarizer · trainingconvergence stabilityHurt
vary the regularization hyperparameter (ablation)autoencoder binarizer · reported experimentsperformance: ~2% variationNo clear effect
use binary instead of reconstructed vectors (ablation)binarized embeddings · document classification and sentiment analysisaccuracyNo clear effect

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