Alexandre Bayle

New paper questions cross-validation stability for model comparison

A new paper published on arXiv demonstrates that cross-validation, a common statistical technique for comparing machine learning models, can produce unstable and invalid inferences. The research specifically highlights …