Research map / Label-efficient and robust learning / Recognition and long-tail
Joint Active Feature Acquisition and Classification with Variable-Size Set Encoding
Hajin Shim, Sung Ju Hwang, Eunho Yang
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 |
|---|---|---|---|
| use feature-level set encoding instead of naive MLP encoding with joint training (ablation) | joint feature acquisition model · Fatty liver (check-up data), uniform cost | AUC: 0.775 -> 0.795 | Mixed |
| use feature-level set encoding instead of naive MLP encoding with joint training (ablation) | joint feature acquisition model · Fatty liver (check-up data), real cost | examination cost (USD): 20.1 -> 13.8 | Improved |
| use feature-level set encoding instead of naive MLP encoding (ablation) | joint model · CUBE-0.3, 20 and 100 features | accuracy and features acquired | Improved |
| train classifier jointly with the agent instead of pre-training it (ablation) | set-encoding model · CUBE | accuracy | Improved |
| train classifier jointly instead of pre-trained (ablation) | set-encoding model · IMT (check-up data), real cost | AUC: 0.760 -> 0.756 (uniform), 0.721 -> 0.750 (real) | Mixed |
| use real feature costs instead of uniform cost (ablation) | joint set-encoding model · Fatty liver (check-up data) | examination cost (USD): 149.8 -> 13.8 | Improved |
| use real feature costs instead of uniform cost (ablation) | joint set-encoding model · IMT (check-up data) | examination cost (USD): 309.8 -> 10.0 | Improved |
| partially share layers between Q network and classifier instead of none or full sharing (ablation) | Enc-Q-C framework · CUBE with 90 dummy features | accuracy and features collected | Improved |
| use joint model with set encoding and real cost instead of MLP using all features | joint set-encoding model · Fatty liver (check-up data) | AUC: 0.812 -> 0.792 | No clear effect |
Closest papers by meaning
- Stochastic Encodings for Active Feature Acquisition (ICML 2025 · similarity 0.87)
- Information Templates: A New Paradigm for Intelligent Active Feature Acquisition (NeurIPS 2026 · similarity 0.82)
- Active Learning with LLMs for Partially Observed and Cost-Aware Scenarios (NeurIPS 2024 · similarity 0.82)
- Provably Neural Active Learning Succeeds via Prioritizing Perplexing Samples (ICML 2024 · similarity 0.82)
- Active feature acquisition via explainability-driven ranking (ICML 2025 · similarity 0.82)
- Universal Rates for Active Learning (NeurIPS 2024 · similarity 0.81)
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.
