Research map / Label-efficient and robust learning / Recognition and long-tail
A Multi-Grained Self-Interpretable Symbolic-Neural Model For Single/Multi-Labeled Text Classification
Xiang Hu, XinYu KONG, Kewei Tu
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add mutual-exclusiveness constraint to the Symbolic-Neural model (ablation) | Fast-R2D2 S.N. (Navigator and others) · CoLA / Navigator / ATIS / SST-2 | accuracy / F1 (Table 1) | Mixed |
| add top-down encoder to the Symbolic-Neural model (ablation) | Fast-R2D2 S.N. · Navigator (Table 1) | F1: +1.57 (89.25 to 90.82 in the Navigator column of Table 1; Fast-R2D2 S.N. at 88.18) | Improved |
| add top-down encoder and mutual-exclusiveness constraint (ablation) | Symbolic-Neural model · sls-movie-eng, sls-movie-trivial, sls-restaurant (span attribution) | F1: +4.78 / +3.53 / +4.45 over base | Improved |
| add top-down encoder alone (ablation) | Symbolic-Neuron · ATIS slot-filling span attribution | F1: -2.44 (35.30 to 32.86) | Hurt |
| add mutual-exclusiveness constraint alone (ablation) | Symbolic-Neuron · ATIS slot-filling span attribution | F1: -3.17 (35.30 to 32.13) | Hurt |
| Symbolic-Neural model vs MIMLL baseline | Fast-R2D2 backbone · sls-movie-trivial / sls-restaurant span attribution | F1 | Mixed |
| Symbolic-Neural model vs IG attribution baseline on long spans | Symbolic-Neuron (span attribution) · sls-movie-eng >5-token bucket | F1: 57.77 vs 34.36 (IG mask) | Improved |
| add top-down encoder to the Symbolic-Neural model (ablation) | Fast-R2D2 S.N. on multi-label tasks · Navigator / Scheduler / Weather (Table 1) | F1 | Mixed |
Closest papers by meaning
- Self-Supervised Learning for Text Recognition: A Critical Survey (IJCV 2025 · similarity 0.85)
- Simultaneous Learning of Trees and Representations for Extreme Classification and Density Estimation (ICML 2017 · similarity 0.85)
- Weakly-Supervised Hierarchical Text Classification (AAAI 2019 · similarity 0.84)
- Jointly Learning to Label Sentences and Tokens (AAAI 2019 · similarity 0.84)
- Harnessing the Power of Beta Scoring in Deep Active Learning for Multi-Label Text Classification (AAAI 2024 · similarity 0.83)
- Liberating Seen Classes: Boosting Few-Shot and Zero-Shot Text Classification via Anchor Generation and Classification Reframing (AAAI 2024 · 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.
