MRA-based Statistical Learning from Incomplete Rankings

Authors: Eric Sibony, Stéphan Clemençon, Jérémie Jakubowicz

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

Reproducibility Variable Result LLM Response
Research Type Experimental Beyond theoretical guarantees, we also provide experimental results that show its statistical performance. Experimental results are also provided in section 6 to illustrate its performance.
Researcher Affiliation Academia Eric Sibony ERIC.SIBONY@TELECOM-PARISTECH.FR St ephan Cl emenc on STEPHAN.CLEMENCON@TELECOM-PARISTECH.FR LTCI UMR No. 5141, Telecom Paris Tech/CNRS, Institut Mines-Telecom, Paris, 75013, France J er emie Jakubowicz JEREMIE.JAKUBOWICZ@TELECOM-SUDPARIS.EDU SAMOVAR UMR No. 5157, Telecom Sud Paris/CNRS, Institut Mines-Telecom, Evry, 91000, France
Pseudocode No The paper describes mathematical frameworks and propositions but does not include any explicitly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any statements or links indicating that its source code is open or publicly available.
Open Datasets Yes and one empirical model, namely the distribution of the 5738 votes in the APA dataset (see Diaconis, 1989) that we consider as a ground truth ranking model.
Dataset Splits No The paper mentions varying the size of the drawn dataset but does not specify any train/validation/test splits or cross-validation setup for reproducing the experiments.
Hardware Specification No The paper does not provide any specific details regarding the hardware (e.g., CPU, GPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions methods like the 'MM algorithm introduced in Hunter (2004)' and the 'estimator from Sun et al. (2012), called SLK', but it does not specify any software dependencies with version numbers.
Experiment Setup Yes In all the experiments, n = 5. For each ranking model, we examine the four different settings where ν is the uniform probability distribution on {A J5K | 2 |A| k} for k = 2, 3, 4, 5, and let the size of the drawn dataset DN vary between 500 and 5000. We then evaluate the performance of an empirical ranking model bq N constructed from DN through a Monte Carlo estimate of the performance E(bq N) averaged from 100 drawings of DN.