Efficient Learning in Polyhedral Games via Best-Response Oracles
Authors: Darshan Chakrabarti, Gabriele Farina, Christian Kroer
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we conduct experiments to demonstrate that our algorithm performs well in practice relative to other projectionfree algorithms when computing Nash and coarse correlated equilibria in polyhedral games. |
| Researcher Affiliation | Academia | 1Columbia University 2MIT dc3595@columbia.edu, gfarina@mit.edu, ck2945@columbia.edu |
| Pseudocode | Yes | Pseudocode for AFW is provided in Appendix A. Algorithm 1: Reflected Gradient OMD with Approximate Proximal Computation (for a generic Player i) |
| Open Source Code | No | No concrete access to source code (specific repository link, explicit code release statement, or code in supplementary materials) for the methodology described in this paper was found. |
| Open Datasets | No | No concrete access information (specific link, DOI, repository name, formal citation with authors/year, or reference to established benchmark datasets with proper attribution) for a publicly available or open dataset was found. The paper refers to 'standard EFG benchmarks' like 'Kuhn poker' and 'Leduc poker' but does not provide details on how to access the corresponding data. |
| Dataset Splits | No | No specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning was found. The paper focuses on online learning in games and discusses concepts like 'averaged iterates' and 'last-iterate convergence' rather than traditional data splits. |
| Hardware Specification | No | No specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments were found. |
| Software Dependencies | No | No specific ancillary software details (e.g., library or solver names with version numbers like Python 3.8, CPLEX 12.4) needed to replicate the experiment were found. |
| Experiment Setup | Yes | For AFW-OMD, AFW-ROMD, FTPL, and OFTPL, we try η 0.01 2[14], where η is the stepsize for our algorithms, while η is the noise used for FTPL and OFTPL. For our algorithms and (O)FTPL, we restrict the number of LMO calls per iteration to be in {1, 2, 3, 4, 5, 10, 20, 100, 200}. |