Research map / Label-efficient and robust learning / Recognition and long-tail
Distributed Negative Sampling for Word Embeddings
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| sample all negatives from a single partition (SNS) instead of per-sample partitions (BNS) (ablation) | SG-SNS · Single-epoch learning time | seconds: 4378 -> 1331 | Helped |
| draw negatives from target word's partition (TNS) instead of SNS (ablation) | SG-TNS · Single-epoch learning time | seconds: 1331 -> 570 | Helped |
| replace reference negative sampling with TNS | SGNS · Single-epoch learning time | speedup: 1 x -> 20.2 x | Helped |
| approximate negative sampling (BNS/SNS/TNS) vs reference | SGNS · Google Analogy | total accuracy %: 71.58 -> 70.03 | Hurt |
| TNS vs SNS accuracy (ablation) | SG-TNS · Google Analogy | total accuracy %: 70.53 -> 70.03 | No clear effect |
Closest papers by meaning
- Spectral Word Embedding with Negative Sampling (AAAI 2018 · similarity 0.87)
- Bayesian Neural Word Embedding (AAAI 2017 · similarity 0.84)
- Closed-Form Training Dynamics Reveal Learned Features and Linear Structure in Word2Vec-like Models (NeurIPS 2025 · similarity 0.84)
- What the Vec? Towards Probabilistically Grounded Embeddings (NeurIPS 2019 · similarity 0.84)
- LightRNN: Memory and Computation-Efficient Recurrent Neural Networks (NeurIPS 2016 · similarity 0.83)
- Generating Enhanced Negatives for Training Language-Based Object Detectors (CVPR 2024 · similarity 0.81)
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.