Acquiring Integer Programs from Data

Authors: Mohit Kumar, Stefano Teso, Luc De Raedt

IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our empirical evaluation shows that ARNOLD can acquire models for a number of realistic benchmark problems. (3) An extensive empirical analysis on a number of integer programs, showing that ARNOLD can acquire good quality programs from a handful of examples.
Researcher Affiliation Academia Mohit Kumar , Stefano Teso and Luc De Raedt KU Leuven {mohit.kumar,stefano.teso,luc.deraedt}@cs.kuleuven.be
Pseudocode Yes Algorithm 1 The ARNOLD search algorithm.
Open Source Code Yes Our code is available at github.com/mohit KULeuven/arnold
Open Datasets Yes To this end, we used ARNOLD for learning 10 satisfaction/satisficing Mini Zinc [Nethercote et al., 2007] benchmark integer programs1, detailed in Table 2. 1From github.com/Mini Zinc/benchmarks and from hakank.org/minizinc.
Dataset Splits Yes Next, we split the dataset into five folds of 25 solutions each and fed ARNOLD with n {1, 2, 10, 25} random solutions from one fold as training set, while using the union of the other four folds for performance evaluation.
Hardware Specification No The paper does not provide specific hardware details such as CPU/GPU models, memory, or cloud instance types used for running experiments. It mentions 'Runtime' but not the underlying hardware.
Software Dependencies No The paper mentions using 'Mini Zinc [Nethercote et al., 2007]' and 'Gecode solver [Schulte et al., 2006]' but does not provide specific version numbers for these or other software dependencies.
Experiment Setup Yes The parameters used were (s, p) = (1, 1), (1, 2), (1, 3), (2, 1), (3, 1), which are large enough to capture the majority of the benchmark problems, and n = 1, 5, 10, 25.