Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Data Agent: Learning to Select Data via End-to-End Dynamic Optimization
Suorong Yang, Fangjian Su, Hai Gan, Ziqi Ye, Jie Li, Baile Xu, Furao Shen, Soujanya Poria
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 difficulty reward (ablation) | Data Agent · Tiny-ImageNet (ResNet-50), 30% | accuracy: 41.5 -> 42.1 | Improved |
| add difficulty reward (ablation) | Data Agent · Tiny-ImageNet (ResNet-50), 50% | accuracy: 42.8 -> 45.0 | Improved |
| add confidence reward (ablation) | Data Agent · Tiny-ImageNet (ResNet-50), 30% | accuracy: 41.5 -> 42.2 | Improved |
| add confidence reward (ablation) | Data Agent · Tiny-ImageNet (ResNet-50), 70% | accuracy: 43.1 -> 48.4 | Improved |
| add confidence reward to difficulty reward (ablation) | Data Agent · Tiny-ImageNet (ResNet-50), 30% | accuracy: 42.1 -> 42.5 | Improved |
| add adaptive reward weighting r (ablation) | Data Agent · Tiny-ImageNet (ResNet-50), 30% | accuracy: 42.5 -> 44.9 | Improved |
| add adaptive reward weighting r (ablation) | Data Agent · Tiny-ImageNet (ResNet-50), 50% | accuracy: 46.1 -> 47.0 | Improved |
| add adaptive reward weighting r (ablation) | Data Agent · Tiny-ImageNet (ResNet-50), 70% | accuracy: 48.6 -> 49.4 | Improved |
| add PPO agent over greedy top-scoring selection by composite reward (ablation) | Data Agent · CIFAR-100 and Tiny-ImageNet | accuracy | Improved |
| use data agent instead of full dataset training | ResNet-50 · ImageNet-1k, 60% selection | accuracy: 76.4 -> 76.8 | Improved |
| use data agent instead of full dataset training | ResNet-50 · ImageNet-R, 60% selection | accuracy: 36.2 -> 38.0 | Improved |
| use data agent instead of full dataset training | ResNet-50 · ImageNet-O, 60% selection | AUPR: 13.2 -> 15.7 | Improved |
Closest papers by meaning
- When Dynamic Data Selection Meets Data Augmentation: Achieving Enhanced Training Acceleration (ICML 2025 · similarity 0.88)
- MAT-Agent: Adaptive Multi-Agent Training Optimization (NeurIPS 2025 · similarity 0.87)
- Adaptive Data Optimization: Dynamic Sample Selection with Scaling Laws (ICLR 2025 · similarity 0.87)
- Scaling Generalist Data-Analytic Agents (ICLR 2026 · similarity 0.85)
- Computational Budget Should Be Considered in Data Selection (NeurIPS 2025 · similarity 0.85)
- A CLIP-Powered Framework for Robust and Generalizable Data Selection (ICLR 2025 · similarity 0.85)
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.
