Research map / Reinforcement learning / Visual and representation RL
Feature Engineering for Predictive Modeling Using Reinforcement Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 RL1 to guide budgeted transformation-graph feature engineering instead of the strongest reported non-RL approach | Random Forest feature engineering · Amazon Employee | FScore: 0.744 -> 0.806 | Helped |
| Use RL1 to guide budgeted transformation-graph feature engineering instead of the strongest reported non-RL approach | Random Forest feature engineering · Bikeshare DC | 1 - relative absolute error: 0.790 -> 0.798 | Helped |
| Use RL1 to guide budgeted transformation-graph feature engineering instead of the strongest reported non-RL approach | Random Forest feature engineering · SpectF | FScore: 0.790 -> 0.788 | Hurt |
| Use RL1 rather than the base dataset without feature engineering | Random Forest · 24 public datasets | FScore or 1 - relative absolute error: 23.8% | Helped |
| Add feature selection as an available transformation (ablation) | Feature engineering system · 48 datasets | final performance gain: 51% | Helped |
| Learn a general transformation bias and condition it on feature types (RL1 rather than RL2) (ablation) | Transformation-graph exploration policy · 10 datasets | exploration efficiency: 4-8 times | Helped |
| Increase the maximum transformation depth from 4 to 5 (ablation) | RL feature engineering · 5 representative datasets | maximum accuracy found | Helped |
| Increase maximum transformation depth to 6 (ablation) | RL feature engineering · 5 representative datasets | maximum accuracy found | Hurt |
Closest papers by meaning
- Reinforcement-Enhanced Autoregressive Feature Transformation: Gradient-steered Search in Continuous Space for Postfix Expressions (NeurIPS 2023 · similarity 0.88)
- Learning MDPs from Features: Predict-Then-Optimize for Sequential Decision Making by Reinforcement Learning (NeurIPS 2021 · similarity 0.85)
- Efficient Exploration and Discriminative World Model Learning with an Object-Centric Abstraction (ICLR 2025 · similarity 0.84)
- Representation-Driven Reinforcement Learning (ICML 2023 · similarity 0.84)
- Program Synthesis Guided Reinforcement Learning for Partially Observed Environments (NeurIPS 2021 · similarity 0.84)
- Is Plug-in Solver Sample-Efficient for Feature-based Reinforcement Learning? (NeurIPS 2020 · 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.