Improving the Efficiency of Dynamic Programming on Tree Decompositions via Machine Learning

Authors: Michael Abseher, Frederico Dusberger, Nysret Musliu, Stefan Woltran

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

Reproducibility Variable Result LLM Response
Research Type Experimental We report on extensive experiments in different problem domains which show a significant speedup when choosing the tree decomposition according to this concept over simply using an arbitrary one of the same width.
Researcher Affiliation Academia Institute of Information Systems 184/2 Vienna University of Technology Favoritenstraße 9 11, 1040 Vienna, Austria
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide concrete access to the source code for the methodology described.
Open Datasets Yes The full training dataset used for our experiments is available under the following link: www.dbai.tuwien.ac.at/research/project/dflat/ features/training.zip
Dataset Splits Yes The results that we present in this paper are obtained with parameter settings that were chosen based on several of our previous experiments using 10-fold cross validation.
Hardware Specification Yes All our experiments were performed on a single core of an AMD Opteron 6308@3.5GHz processor running Debian GNU/Linux 7 (kernel 3.2.0-4-amd64)
Software Dependencies Yes We evaluate our approach using two recently developed DP solvers, D-FLAT (v. 1.0.1) and SEQUOIA (v. 0.9). The subsequent machine learning tasks were carried out with WEKA 3.6.11 [Hall et al., 2009].
Experiment Setup No The paper mentions that parameter settings were chosen based on previous experiments and 10-fold cross-validation, but it does not provide specific hyperparameter values or detailed system-level training configurations for the models used (e.g., learning rates, batch sizes, optimizer details).