Every few days, we will be publishing lay abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.
The article featured today is from Stat and the full article, published in the Deep Learning from Statistical Perspectives special issue and in issue 8.1, is available to read online here.
Liu, Y, Zheng, C. Deep latent variable models for generating knockoffs. Stat. 2019; 8:e260. doi: 10.1002/sta4.260
Machine learning models, especially Deep Learning models are often considered as ‘Black Box’ predictors. Although there is great interest in applying these methods in important science/medical problems, the lack of interpretability and statistical inference become the main criticism. Our paper aims at breaking the dilemma by using deep learning to equip machine learning methods with the ability of ‘variable selection’ and provide statistical inference by controlling for False Discovery Rate. This goal is realized through generating knockoffs of the potential features and implementing Model-X variable selection schemes. In this paper, we provide a knockoff generating methods with variational auto-encoder. It is ready to be implemented by our released python code for generating knockoffs with fast Variational Bayesian algorithm.