Research map / Label-efficient and robust learning / Recognition and long-tail
Neural Bag-of-Ngrams
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 |
|---|---|---|---|
| add bi-grams to uni-grams (ablation) | LGNR · MR | accuracy: 77.92 -> 78.89 | Helped |
| add tri-grams on top of bi-grams (ablation) | TGNR · MR (small dataset) | accuracy: 69.79 -> 63.25 | Hurt |
| add tri-grams on top of bi-grams (ablation) | TGNR · IMDB (large vocabulary) | accuracy: 91.64 -> 91.83 | Helped |
| use text guided (TGNR) instead of context guided (CGNR) n-gram representation (ablation) | Neural-BoN uni-gram · IMDB | accuracy: 84.06 -> 90.24 | Helped |
| use label guided (LGNR) instead of TGNR (ablation) | Neural-BoN bi-gram · RT-2k | accuracy: 88.75 -> 86.5 | Hurt |
| train with unlabeled IMDB corpus (+corpus) (ablation) | TGNR · MR | accuracy: 69.79 -> 79.25 (+9.46) | Helped |
| train with unlabeled IMDB corpus (+corpus) (ablation) | CGNR · IMDB | accuracy: 85.35 -> 86(+0.65) | Helped |
| add Naive Bayes weighting (ablation) | CGNR · AthR | accuracy: 76.72 -> 78.26 (+1.54) | Helped |
| add Naive Bayes weighting (ablation) | TGNR · CR | accuracy: 77.19 -> 77.72 (+0.53) | No clear effect |
| ensemble with one-hot BoN representation (ablation) | TGNR · AthR | accuracy: 84.01 -> 88.35 (+3.64) | Helped |
| ensemble with one-hot BoN representation (ablation) | LGNR · MR | accuracy: 78.89 -> 79.55 (+0.66) | No clear effect |
| train with non-sentiment corpora (STATMT NEWS, Wikipedia) (ablation) | Neural-BoN · text classification | accuracy | No clear effect |
Closest papers by meaning
- A New Method of Region Embedding for Text Classification (ICLR 2018 · similarity 0.86)
- Character n-Gram Embeddings to Improve RNN Language Models (AAAI 2019 · similarity 0.86)
- A Simple but Tough-to-Beat Baseline for Sentence Embeddings (ICLR 2017 · similarity 0.85)
- A Compressed Sensing View of Unsupervised Text Embeddings, Bag-of-n-Grams, and LSTMs (ICLR 2018 · similarity 0.84)
- Learning Semantic Representations for Novel Words: Leveraging Both Form and Context (AAAI 2019 · similarity 0.84)
- NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models (ICLR 2025 · similarity 0.83)
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.