Research map / Label-efficient and robust learning / Recognition and long-tail
Human-in-the-Loop Feature Selection
Alvaro H. C. Correia, Freddy Lecue
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 simulated human feature feedback (SF, cosine) (ablation) | per-example feature selection · Cluttered MNIST 60x60 | accuracy (%): 85.35 -> 92.30 | Improved |
| add real expert feature feedback (SF, cosine) (ablation) | per-example feature selection · Project Risk Classification | accuracy (%): 29.53 -> 82.49 | Improved |
| use SF instead of PD estimator (ablation) | feedback-based selection · Project Risk Classification | accuracy (%): 77.51 -> 82.49 | Improved |
| use SF instead of PD estimator (ablation) | feedback-based selection · Cluttered MNIST 60x60 | accuracy (%): 88.40 -> 92.30 | Improved |
| set Gumbel-softmax temperature to 1.0 vs other values (ablation) | PD estimator · Project Risk Classification | accuracy (%): 77.82 -> 81.87 | Improved |
| increase number of states K from 2 to 8 (ablation) | PD estimator · Project Risk Classification | accuracy (%): 73.11 -> 79.44 | Mixed |
| increase number of states K from 2 to 20 (ablation) | PD estimator · Cluttered MNIST | accuracy (%): 89.50 -> 85.40 | Hurt |
| start feedback later (epoch 40 vs 80) (ablation) | SF with MSE feedback · Project Risk Classification | accuracy (%): 80.11 -> 52.10 | Hurt |
| remove all conflicting feedback (ablation) | SF with MSE feedback · Project Risk Classification | accuracy (%): 86.67 -> 69.87 | Hurt |
| increase feedback amount from 50% to 100% (ablation) | SF with MSE feedback · Project Risk Classification | accuracy (%): 61.65 -> 80.11 | Improved |
| MSE vs cosine dissimilarity (ablation) | feedback loss · Project Risk Classification | accuracy (%): 82.49 vs 80.11 | No clear effect |
Closest papers by meaning
- L-WISE: Boosting Human Visual Category Learning Through Model-Based Image Selection and Enhancement (ICLR 2025 · similarity 0.85)
- Stochastic Encodings for Active Feature Acquisition (ICML 2025 · similarity 0.85)
- Leveraging Sub-Optimal Data for Human-in-the-Loop Reinforcement Learning (ICLR 2025 · similarity 0.84)
- Estimating Conditional Mutual Information for Dynamic Feature Selection (ICLR 2024 · similarity 0.84)
- SAND: One-Shot Feature Selection with Additive Noise Distortion (ICML 2025 · similarity 0.84)
- Capturing Individual Human Preferences with Reward Features (NeurIPS 2025 · similarity 0.84)
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.
