Research map / Label-efficient and robust learning / Recognition and long-tail
Merging Statistical Feature via Adaptive Gate for Improved Text Classification
Xianming Li, Zongxi Li, Haoran Xie, Qing Li
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add corpus-level statistical features using the adaptive gate | Bert-base Uncased · Subj | accuracy: 97.03 -> 97.89 | Improved |
| add corpus-level statistical features using the adaptive gate | Bert-base Uncased · SST-1 | accuracy: 53.78 -> 55.72 | Improved |
| add corpus-level statistical features using the adaptive gate | Bert-base Uncased · SST-2 | accuracy: 91.07 -> 93.27 | Improved |
| add corpus-level statistical features using the adaptive gate | Bert-base Uncased · TREC | accuracy: 96.48 -> 97.65 | Improved |
| add corpus-level statistical features using the adaptive gate | Bert-base Uncased · AG’s News | accuracy: 93.30 -> 93.82 | Improved |
| add corpus-level statistical features using the adaptive gate | Bert-base Uncased · Yelp F. | accuracy: 95.73 -> 96.79 | Improved |
| add corpus-level statistical features using the adaptive gate | Bert-base Uncased · Yelp P. | accuracy: 68.46 -> 70.74 | Improved |
| use a VAE rather than a vanilla autoencoder to encode TCoL (ablation) | CNN+AGN · SST-2, TREC, Yelp F., and Yelp P. | accuracy: 81.98 -> 82.24; 92.02 -> 92.11; 94.89 -> 95.15; 67.11 -> 67.93 | Improved |
| adaptively filter corrupted statistical information rather than accept all of it (ϵ=0.2 vs ϵ=0.5) (ablation) | CNN+AGN · SST-2 and TREC with shuffled TCoL | accuracy: 78.25 -> 80.05; 87.69 -> 89.64 | Improved |
| adaptively use only some statistical information instead of rejecting all or accepting all (ablation) | AGN · SST-2 and TREC | accuracy | Improved |
Closest papers by meaning
- Graph Convolutional Networks for Text Classification (AAAI 2019 · similarity 0.83)
- PromptBoosting: Black-Box Text Classification with Ten Forward Passes (ICML 2023 · similarity 0.83)
- Active Discriminative Text Representation Learning (AAAI 2017 · similarity 0.83)
- TACIT: A Target-Agnostic Feature Disentanglement Framework for Cross-Domain Text Classification (AAAI 2024 · similarity 0.83)
- Fooling Explanations in Text Classifiers (ICLR 2022 · similarity 0.83)
- Mixture of Expert/Imitator Networks: Scalable Semi-Supervised Learning Framework (AAAI 2019 · 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.
