A low variance consistent test of relative dependency
Authors: Wacha Bounliphone, Arthur Gretton, Arthur Tenenhaus, Matthew Blaschko
ICML 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The effectiveness of the test is demonstrated on several real-world problems: we identify language groups from a multilingual corpus, and we prove that tumor location is more dependent on gene expression than chromosomal imbalances. and 5. Experiments We apply our estimates of statistical dependence to three challenging problems. |
| Researcher Affiliation | Academia | Wacha Bounliphone WACHA.BOUNLIPHONE@CENTRALESUPELEC.FR Centrale Sup elec & Inria, Grande Voie des Vignes, 92295 Chˆatenay-Malabry, France Arthur Gretton ARTHUR.GRETTON@GMAIL.COM Gatsby Computational Neuroscience Unit, University College London, United Kingdom Arthur Tenenhaus ARTHUR.TENENHAUS@CENTRALESUPELEC.FR Centrale Sup elec, 3 rue Joliot-Curie, 91192 Gif-Sur-Yvette, France Matthew B. Blaschko MATTHEW.BLASCHKO@INRIA.FR Inria & Centrale Sup elec, Grande Voie des Vignes, 92295 Chˆatenay-Malabry, France |
| Pseudocode | Yes | Algorithm 1 Successive rotation for generalized highdimensional relative tests of dependency (cf. Section 4) |
| Open Source Code | Yes | Source code is available for download at https://github. com/wbounliphone/reldep. |
| Open Datasets | Yes | We use a real world dataset taken from the parallel European Parliament corpus (Koehn, 2005). and pre-treatment frozen tumor samples were obtained from 53 children with newly diagnosed p HGG from Necker Enfants Malades (Paris, France) from Puget et al, (2012). |
| Dataset Splits | No | The paper describes a theoretical splitting of samples for a simpler test ('split the samples from Px into two equal sized sets denoted by X and X') but does not provide specific train/validation/test splits for the empirical experiments performed in Section 5. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU models, CPU types, or memory specifications used for running the experiments. |
| Software Dependencies | No | The paper mentions software tools like NLTK and Snowball stemmer, and concepts like TF-IDF, but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | Yes | used a Gaussian kernel with bandwidth σ selected as the median pairwise distance between data points. and For pre-processing, we removed stop-words (http:// www.nltk.org) and performed stemming (http:// snowball.tartarus.org). We applied the TF-IDF model as a feature representation and used a Gaussian kernel with the bandwidth σ set per language as the median pairwise distance between documents. and For X, we use a linear kernel, which is characteristic for indicator variables, and for Y and Z, the kernel was chosen to be the Gaussian kernel with σ selected as the median of pairwise distances. |