Learning Weighted Model Integration Distributions

Authors: Paolo Morettin, Samuel Kolb, Stefano Teso, Andrea Passerini5224-5231

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experimental results with synthetic and real-world data highlight the promise of the approach.
Researcher Affiliation Academia 1University of Trento, Italy {paolo.morettin, stefano.teso, andrea.passerini}@unitn.it 2KU Leuven, Belgium samuel.kolb@cs.kuleuven.be
Pseudocode Yes Algorithm 1 The inner loop of the INCAL+ algorithm.
Open Source Code No The code is available at: URL ANONYMIZED
Open Datasets Yes We evaluated LARIAT on the hybrid UCI benchmarks contained in the MLC++ library, which includes 18 hybrid datasets from different real-world domains.
Dataset Splits Yes For each configuration, we generated 20 different ground-truth models and relative dataset, each consisting of 500 training and 50 validation examples.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact CPU/GPU models, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper mentions using 'Math SAT solver' and 'Z3' but does not provide specific version numbers for these software dependencies used in their experiments.
Experiment Setup Yes we implemented INCAL+ using the Math SAT solver, applied it to the synthetic datasets with a timeout of 300 seconds for each call, and measured the misclassification error between the true and the learned support.