A Coreset Learning Reality Check

Subsampling algorithms are a natural approach to reduce data size before fitting models on massive datasets. In recent years, several works have proposed methods for subsampling rows from matrix while maintaining relevant information classification. While these supported by theory and limited experiments, date there has not been comprehensive evaluation of methods. our work, we directly compare multiple logistic regression drawn the coreset optimal literature discover inconsistencies in their effectiveness. many cases, do outperform simple uniform subsampling.