Conflict Directed Clause Learning for Maximum Weighted Clique Problem

Authors: Emmanuel Hebrard, George Katsirelos

IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on several standard benchmarks show that this approach is very promising and compares favourably with state-of-the-art methods on many families of instances. 9 Experimental Evaluation We implemented our approach in C++, denoted cdcl, on top of the MINICSP solver2. We compared it with the solvers mwclq [Fang et al., 2016], wlmc [Jiang et al., 2017], cliquer [ Osterg ard, 2001], OTClique [Shimizu et al., 2017] and an implementation of Tavares method by the authors of [Mc Creesh et al., 2017]. We used the benchmarks introduced in [Mc Creesh et al., 2017] divided in four classes
Researcher Affiliation Academia Emmanuel Hebrard1 and George Katsirelos2 1 LAAS-CNRS, Universit e de Toulouse, CNRS, Toulouse, France 2 MIAT, INRA, Toulouse, France hebrard@laas.fr, gkatsi@gmail.com
Pseudocode Yes Algorithm 1: WEIGHTEDIS(G = (V, E), w, k, A) Algorithm 2: DUALBOUND(G = (V, E)) Algorithm 3: MCCPROPAGATE(G = (V, E), M, lb, ub) Algorithm 4: EXPLAINBOUND(G, w, A, M, L)
Open Source Code Yes We implemented our approach in C++, denoted cdcl, on top of the MINICSP solver2. 2Sources available at: https://bitbucket.org/gkatsi/minicsp
Open Datasets Yes We used the benchmarks introduced in [Mc Creesh et al., 2017] divided in four classes: WDP Winner Determination Problem... REF Research Excellence Network... EC-CODE Error-correcting Codes... KIDNEY Kidney Exchange... Moreover, we also provide results on the classic DIMACS and BHOSLIB benchmarks.
Dataset Splits No The paper uses standard benchmark instances, but does not provide details on specific training, validation, or test dataset splits, as is common for machine learning tasks. The evaluation is performed on entire instances.
Hardware Specification Yes Every method was run with a time limit of 1h and a memory limit of 3.5GB3 on a cluster with 4 nodes, each with 35 Intel Xeon E5-2695 2.10GHz cores running Ubuntu Linux 16.04.4.
Software Dependencies No The paper mentions implementation in C++ and on top of the MINICSP solver, but does not specify their versions. It also mentions 'Ubuntu Linux 16.04.4' as the operating system, but no specific versions for other software dependencies.
Experiment Setup Yes Every method was run with a time limit of 1h and a memory limit of 3.5GB. all the results that we report for our method in section 9 were obtained using VSIDS for up to 50000 nodes, and then switching to branching on v minimising |C(v)|/w(v).