Distributed Multiplicative Weights Methods for DCOP

Authors: Daisuke Hatano, Yuichi Yoshida

AAAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We experimentally demonstrate that our methods have good scalability, and in particular, the second method outperforms existing algorithms in terms of solution quality and efficiency.
Researcher Affiliation Collaboration Daisuke Hatano National Institute of Informatics JST, ERATO, Kawarabayashi Large Graph Project hatano@nii.ac.jp Yuichi Yoshida National Institute of Informatics, and Preferred Infrastructure, Inc. yyoshida@nii.ac.jp
Pseudocode Yes Algorithm 1 The multiplicative weights method; Algorithm 2 Computing subgradients; Algorithm 3 DMW-LP (with the majority strategy); Algorithm 4 DMW-Game (with the majority strategy)
Open Source Code No The paper does not include any explicit statement about making its own source code available or provide a link to a code repository. It mentions using 'FRODO version 2.11' which is a third-party tool.
Open Datasets No The paper describes generating its own DCOP instances (random, scale-free, meeting scheduling problems using FRODO's instance generator) but does not provide concrete access (link, DOI, specific citation to an existing public dataset used directly) to the *specific* datasets generated or used for the experiments.
Dataset Splits No The paper does not explicitly provide training/validation/test dataset splits (e.g., percentages, sample counts, or references to predefined splits).
Hardware Specification Yes We conducted experiments on an Ubuntu server with Intel Core-i7 3770@3.4GHz and 4GB of memory.
Software Dependencies Yes For Max Sum, DSA and MGM, we used the code in FRODO version 2.11 (L eaut e, Ottens, and Szymanek 2009) with the default setting.
Experiment Setup Yes For DMW-LP and DMW-Game, we set η to be 0.04 and 0.5, respectively.