Research map / Multi-agent systems and games / Games and equilibria
A Mean-Field Game Approach to Cloud Resource Management with Function Approximation
Weichao Mao, Haoran Qiu, Chen Wang, Hubertus Franke, Zbigniew Kalbarczyk, Ravi Iyer, Tamer Basar
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 |
|---|---|---|---|
| replace linear function approximation with a two-layer neural network in NAC (ablation) | NAC · OpenWhisk training reward | reward after convergence: 18.8% higher | Improved |
| replace linear function approximation with a neural network in NAC (ablation) | NAC · OpenWhisk training | episodes to converge: about 300 fewer episodes | Improved |
| use NAC-NN instead of OpenWhisk original resource manager | OpenWhisk resource management · function serving | p99 latency: 33.6% to 67.3% lower | Improved |
| use NAC-NN instead of ENSURE heuristic | OpenWhisk resource management · function serving | p99 latency: 14.8% to 29.6% lower | Improved |
| use NAC-NN instead of ENSURE | OpenWhisk resource management · function serving | containers in use: 29% fewer | Improved |
| use NAC-NN instead of OpenWhisk original | OpenWhisk resource management · function serving | containers in use: 37% fewer | Improved |
| use NAC-Linear instead of ENSURE | OpenWhisk resource management · function serving | p99 latency: up to 25.5% | Improved |
| use NAC-Linear instead of ENSURE | OpenWhisk resource management · function serving | containers in use: up to 24% fewer | Improved |
| replace double-loop fixed-point iteration (DL-NAC) with single-loop online NAC | mean-field game solver · SIS and LQ MFG | exploitability convergence | Improved |
Closest papers by meaning
- Solving Continuous Mean Field Games: Deep Reinforcement Learning for Non-Stationary Dynamics (NeurIPS 2025 · similarity 0.86)
- Transition-Informed Reinforcement Learning for Large-Scale Stackelberg Mean-Field Games (AAAI 2024 · similarity 0.85)
- Finite-Sample Convergence Bounds for Trust Region Policy Optimization in Mean Field Games (ICML 2025 · similarity 0.85)
- Bench-MFG: A Benchmark Suite for Learning in Stationary Mean Field Games (NeurIPS 2026 · similarity 0.85)
- Stochastic Semi-Gradient Descent for Learning Mean Field Games with Population-Aware Function Approximation (ICLR 2025 · similarity 0.84)
- Model-Based RL for Mean-Field Games is not Statistically Harder than Single-Agent RL (ICML 2024 · 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.
