Familywise Error Rate Control by Interactive Unmasking
Authors: Boyan Duan, Aaditya Ramdas, Larry Wasserman
ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate via numerical experiments the power of our test under structured non-nulls, and then explore new forms of masking. |
| Researcher Affiliation | Academia | 1Department of Statistics and Data Science, Carnegie Mellon University 2Machine Learning Department, Carnegie Mellon University. Correspondence to: Boyan Duan <boyand@stat.cmu.edu>. |
| Pseudocode | Yes | Algorithm 1 The i-FWER test |
| Open Source Code | Yes | Code can be found in https://github.com/duanby/i-FWER. |
| Open Datasets | Yes | The airway data is an RNA-Seq dataset targeting the identification of differentially expressed genes in airway smooth muscle cell lines in response to dexamethasone, which contains 33469 genes (hypotheses) and a univariate covariate (the logarithm of normalized sample size) for each gene. ... Data in Section 5 is collected by (Himes et al., 2014) and available in R package airway. |
| Dataset Splits | No | The paper describes simulation settings and experiments but does not specify train, validation, and test splits for datasets in the conventional machine learning sense. The experiments involve statistical testing and simulations, not typical dataset partitioning for model training/evaluation. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware used for running the experiments, such as CPU or GPU models. |
| Software Dependencies | Yes | It was tested on mac OS using R (version 3.6.0) and the following packages: magrittr, splines, robustbase, ggplot2. ... We follow (Ignatiadis et al., 2016) and (Lei & Fithian, 2018) to analyze the data using DEseq2 package (Love et al., 2014). |
| Experiment Setup | Yes | For all experiments in the paper, the FWER control is set at level α = 0.2, and the power is averaged over 500 repetitions. ... Given two parameters d and δ (eg. d = 5, δ = 5%). ... masking parameter p chosen as α/20, α/10, α/2 |