Research map / Learning theory and algorithms / Combinatorial and search optimization
Symbolic Regression via Deep Reinforcement Learning Enhanced Genetic Programming Seeding
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 |
|---|---|---|---|
| add GP with RNN-seeded populations to RNN-only (DSR) (ablation) | Ours · Nguyen+R+Livermore (all) | recovery rate: 45.19 -> 74.92 | Helped |
| seed GP with RNN samples instead of random seeds (ablation) | Ours · Nguyen+R+Livermore (all) | recovery rate: 63.57 -> 74.92 | Helped |
| train RNN on GP samples (off-policy) instead of only RNN samples (ablation) | Ours · Nguyen+R+Livermore (all) | recovery rate: 66.27 -> 74.92 | Helped |
| add RL (RNN) samples to training in addition to GP samples (ablation) | Ours · Nguyen+R+Livermore (all) | recovery rate: 74.81 -> 74.92 | No clear effect |
| use PQT instead of VPG/RSPG trainer (ablation) | Ours · Nguyen+R+Livermore (all) | recovery rate: 74.27 -> 74.92 | No clear effect |
| add entropy regularizer (ablation) | Ours · Nguyen+R+Livermore (all) | recovery rate: 73.95 -> 74.92 | No clear effect |
| use mixed mutation types instead of uniform only (ablation) | Ours · Nguyen+R+Livermore (all) | recovery rate: 72.65 -> 74.92 | Helped |
| add inverse/trig constraints (ablation) | Ours · Nguyen+R+Livermore (all) | recovery rate: 71.35 -> 74.92 | Helped |
| train RNN (nonzero learning rate) (ablation) | Ours · Nguyen+R+Livermore (all) | recovery rate: 65.95 -> 74.92 | Helped |
| hybrid method versus DSR on min NMSE for unsolved benchmarks | Ours · Nguyen-12 | min NMSE: 2.01 × 10−2 -> 1.41 × 10−4 | Helped |
Closest papers by meaning
- Racing Control Variable Genetic Programming for Symbolic Regression (AAAI 2024 · similarity 0.89)
- SYMBOL: Generating Flexible Black-Box Optimizers through Symbolic Equation Learning (ICLR 2024 · similarity 0.88)
- Improving Monte Carlo Tree Search for Symbolic Regression (NeurIPS 2025 · similarity 0.88)
- DisCo-DSO: Coupling Discrete and Continuous Optimization for Efficient Generative Design in Hybrid Spaces (AAAI 2025 · similarity 0.87)
- MetaSymNet: A Tree-like Symbol Network with Adaptive Architecture and Activation Functions (AAAI 2025 · similarity 0.87)
- Symbolic Regression with a Learned Concept Library (NeurIPS 2024 · similarity 0.87)
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.