Research map / Image generation / GANs
DefogGAN: Predicting Hidden Information in the StarCraft Fog of War with Generative Adversarial Nets
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 DefogGAN's predictive generator to accumulated-observation baseline | StarCraft defogging · StarCraft test set | MSE: 0.00370 -> 0.00208 | Helped |
| Add DefogGAN's predictive generator to accumulated-observation baseline | StarCraft defogging · StarCraft test set | F1: 0.850 -> 0.856 | No clear effect |
| Use DefogGAN instead of CED | StarCraft defogging · StarCraft test set | F1: 0.682 -> 0.856 | Helped |
| Add current partial observation (ablation) | DefogGAN · StarCraft test set | MSE: 0.00293 -> 0.00208 | Helped |
| Add accumulated past partial observation (ablation) | DefogGAN · StarCraft test set | MSE: 0.00426 -> 0.00208 | Helped |
| Add adversarial loss (ablation) | DefogGAN · StarCraft test set | F1: 0.662 -> 0.856 | Helped |
| Add reconstruction loss (ablation) | DefogGAN · StarCraft test set | MSE: 1.73986 -> 0.00208 | Helped |
| Use pyramidal reconstruction loss instead of L2 loss (ablation) | DefogGAN · StarCraft test set | MSE: 0.00210 -> 0.00208 | Helped |
| Add observation-preserving connection (ablation) | DefogGAN · StarCraft test set | F1: 0.516 -> 0.856 | Helped |
| Train with different amounts of replay frames (ablation) | DefogGAN · StarCraft test set, 5s–600s training intervals | MSE: 0.00211; 0.00208; 0.00215; 0.00213; 0.00230 | Mixed |
Closest papers by meaning
- Forward Modeling for Partial Observation Strategy Games - A StarCraft Defogger (NeurIPS 2018 · similarity 0.91)
- StarCraft Motion: A Dataset for Agent Simulation in Adversarial and Partially Observable Scenarios (NeurIPS 2026 · similarity 0.86)
- Learning to Simulate Dynamic Environments With GameGAN (CVPR 2020 · similarity 0.84)
- Diffusion Models Are Real-Time Game Engines (ICLR 2025 · similarity 0.82)
- Generative Adversarial Equilibrium Solvers (ICLR 2024 · similarity 0.82)
- Expected flow networks in stochastic environments and two-player zero-sum games (ICLR 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.