Beyond Strict Competition: Approximate Convergence of Multi-agent Q-Learning Dynamics
Authors: Aamal Hussain, Francesco Belardinelli, Georgios Piliouras
IJCAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | As our experiments show, these guarantees are independent of whether the dynamics ultimately reach an equilibrium, or remain non-convergent. Also, section 4 is titled "Experiments on Near NZSG". |
| Researcher Affiliation | Academia | Aamal Hussain1 , Francesco Belardinelli1 , Georgios Piliouras2 1Imperial College London 2Singapore University of Technology and Design {aamal.hussain15, francesco.belardinelli}@imperial.ac.uk, georgios@sutd.edu.sg |
| Pseudocode | No | The paper does not contain any pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any explicit statements about open-source code availability or links to code repositories. |
| Open Datasets | No | The paper states that they "generate a two-action, zero-sum network game" and "perturb the payoff matrices to generate five near zero-sum games." This indicates they generated their own data rather than using a publicly available dataset. |
| Dataset Splits | No | The paper does not explicitly provide training/test/validation dataset splits. They generate game instances and observe dynamics, but do not mention specific percentages or sample counts for data partitioning. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used to run the experiments, such as CPU/GPU models or memory specifications. |
| Software Dependencies | No | The paper does not provide specific version numbers for any software dependencies or libraries used in the experiments. |
| Experiment Setup | Yes | In all cases, T = 0.75. Then, we perturb the payoff matrices to generate five near zero-sum games. When examining the effect of noise, we take the same network game setup and periodically (every 50 iterations) add noise to the payoff matrices. Figure 4: after 1 x 10^6 iterations. |