Learning Mixtures of Random Utility Models

Authors: Zhibing Zhao, Tristan Villamil, Lirong Xia

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on synthetic data show that the sandwich algorithm achieves the highest statistical efficiency and GMM is the most computationally efficient. Experiments on real-world data at Preflib show that Gaussian k-RUMs provide better fitness than a single Gaussian RUM, the Plackett-Luce model, and mixtures of Plackett-Luce models w.r.t. commonly-used model fitness criteria.
Researcher Affiliation Academia Zhibing Zhao, Tristan Villamil, Lirong Xia Rensselaer Polytechnic Institute 110 8th Street, Troy, NY, USA {zhaoz6, villat2}@rpi.edu, xial@cs.rpi.edu
Pseudocode Yes Algorithm 1 E-GMM Algorithm Input: Profile P of n rankings, the number of components k, the number of iterations T. Output: α(T +1) r , θ(r,T +1), where r = 1, 2, , k.
Open Source Code No The paper does not provide an explicit statement or link indicating that the source code for their methodology is openly available.
Open Datasets Yes Experiments on real-world Preflib data (Mattei and Walsh 2013)
Dataset Splits No The paper does not specify the exact training, validation, and test dataset splits (e.g., percentages, absolute counts, or explicit standard splits).
Hardware Specification Yes All experiments were run on an Ubuntu Linux server with Intel Xeon E5 v3 CPUs clocked at 3.50 GHz.
Software Dependencies No The paper states 'We implemented all algorithms with Matlab' but does not provide a specific version number for Matlab or any other key software libraries used with version numbers.
Experiment Setup No While the paper mentions '10 EM iterations' and '5 EM iterations' in figure captions, it does not provide a comprehensive list of hyperparameters, optimizer settings, or other detailed experimental setup configurations in the main text.