Research map / Label-efficient and robust learning / Recognition and long-tail
Table2Analysis: Modeling and Recommendation of Common Analysis Patterns for Multi-Dimensional Data
Mengyu Zhou, Wang Tao, Ji Pengxin, Han Shi, Zhang Dongmei
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 Search Sampling on top of pretrained model (ablation) | Table2Analysis DQN · PivotTable test set | R@1: 0.3108 -> 0.6518 | Improved |
| baseline scheduled sampling (DAD/MIXER-style) after pretraining | Table2Analysis DQN · PivotTable test set | R@1: 0.3108 -> 0.0302 | Hurt |
| use large model instead of small (ablation) | Table2Analysis DQN · DQN pretrain validation | F1: 0.8415 -> 0.8939 | Improved |
| remove semantic embeddings (ablation) | Table2Analysis DQN · DQN pretrain validation | F1 (large): 0.8939 -> 0.8533 | Hurt |
| remove data statistic features (ablation) | Table2Analysis DQN · DQN pretrain validation | F1 (large): 0.8939 -> 0.8882 | Hurt |
| change loss class weight from (0.8,1) to (0.08,1) (ablation) | Table2Analysis DQN · DQN pretrain validation | F1 (large): 0.8939 -> 0.8086 | Hurt |
| OU-noise exploration instead of blind explore (ablation) | Search Sampling · validation R@k | R@k | Improved |
| start Search Sampling from pretrained model instead of random init (ablation) | Search Sampling · validation R@k | R@k: 0.4823 vs 0.0121 at test R@5 (pretrain vs random) | Improved |
| class weight 0.2 vs 0.8 during Search Sampling (ablation) | Search Sampling · validation R@k | R@k | Improved |
Closest papers by meaning
- Text2Analysis: A Benchmark of Table Question Answering with Advanced Data Analysis and Unclear Queries (AAAI 2024 · similarity 0.84)
- MMTU: A Massive Multi-Task Table Understanding and Reasoning Benchmark (NeurIPS 2025 · similarity 0.84)
- WikiDBs: A Large-Scale Corpus Of Relational Databases From Wikidata (NeurIPS 2024 · similarity 0.83)
- UniTabE: A Universal Pretraining Protocol for Tabular Foundation Model in Data Science (ICLR 2024 · similarity 0.82)
- ClavaDDPM: Multi-relational Data Synthesis with Cluster-guided Diffusion Models (NeurIPS 2024 · similarity 0.82)
- TableRAG: Million-Token Table Understanding with Language Models (NeurIPS 2024 · similarity 0.82)
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.
