Research map / Reinforcement learning / Policy gradient and actor-critic
Policy Learning for Continuous Space Security Games Using Neural Networks
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 fictitious play (replay memory) to policy gradient (ablation) | OptGradFP · Forest game single state | opponent best response utility: 481.14 -> 398.2 | Helped |
| replace best-response opponent updates (StackGrad) with soft policy gradient plus fictitious play | defender policy learning · Forest game single state | opponent best response utility: 596.0 -> 398.2 | Helped |
| OptGradFP vs StackGradFP (policy-gradient opponent vs best-response opponent, both with FP) | defender policy learning · Forest game single state | opponent best response utility: 399.4 -> 398.2 | No clear effect |
| replace opponent best response computation with policy gradient step | StackGradFP · Forest game single state | computation time (s): 18426.5 -> 12257.6 | Helped |
| add fictitious play replay memory (ablation) | OptGrad · Forest game single state | computation time (s): 3522.9 -> 12257.6 | Hurt |
| use CNN-predicted strategy on unseen states instead of per-state computed strategy | OptGradFP-NN · Forest game 10 unseen test states | opponent best response utility: 414.4 -> 375.6 (state 0); 400.0 -> 297.5 (state 5) | Hurt |
Closest papers by meaning
- Deep Reinforcement Learning in Continuous Action Spaces: a Case Study in the Game of Simulated Curling (ICML 2018 · similarity 0.86)
- Deep Reinforcement Learning for Green Security Games with Real-Time Information (AAAI 2019 · similarity 0.85)
- NSGZero: Efficiently Learning Non-exploitable Policy in Large-Scale Network Security Games with Neural Monte Carlo Tree Search (AAAI 2022 · similarity 0.85)
- Learning to Stop: Deep Learning for Mean Field Optimal Stopping (ICML 2025 · similarity 0.83)
- Solving Continuous Mean Field Games: Deep Reinforcement Learning for Non-Stationary Dynamics (NeurIPS 2025 · similarity 0.83)
- Identifying Policy Gradient Subspaces (ICLR 2024 · similarity 0.82)
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.