Diverse Randomized Agents Vote to Win

Authors: Albert Jiang, Leandro Soriano Marcolino, Ariel D Procaccia, Tuomas Sandholm, Nisarg Shah, Milind Tambe

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments, which pit teams of computer Go agents against strong agents, provide evidence for the effectiveness of voting when agents are diverse.
Researcher Affiliation Academia Albert Xin Jiang Trinity University xjiang@trinity.edu Leandro Soriano Marcolino USC sorianom@usc.edu Ariel D. Procaccia CMU arielpro@cs.cmu.edu Tuomas Sandholm CMU sandholm@cs.cmu.edu Nisarg Shah CMU nkshah@cs.cmu.edu Milind Tambe USC tambe@usc.edu
Pseudocode No No pseudocode or algorithm blocks are present in the paper.
Open Source Code No The paper mentions using "Fuego 1.1", described as "open source, publicly available Go program", and states they "modified Fuego". However, it does not explicitly state that their modifications or the code used for their methodology is open source or available.
Open Datasets No The paper mentions simulating Go games and using the Fuego program, but does not provide access information (link, DOI, citation for a public dataset) for any experimental data.
Dataset Splits No The paper describes simulating Go games and comparing winning rates, which is not based on traditional dataset splits (training, validation, test) in the way a machine learning dataset would be.
Hardware Specification Yes All results were obtained by simulating 1000 9 9 Go games, in an HP dl165 with dual dodeca core, 2.33GHz processors and 48GB of RAM.
Software Dependencies No The paper mentions "Fuego 1.1" as the primary Go program used, but does not provide specific version numbers for any ancillary software libraries, frameworks, or dependencies.
Experiment Setup Yes We sample random values for a set of parameters for each generated agent, in order to change its behavior. In Appendix G we list the sampled parameters, and the range of sampled values.