Classic Paper: Predicting Algorithm Runtime with Machine Learning by Hutter et al. (2014)

tomssilver · x · 2026-08-03

This week's recommended paper is 'Algorithm Runtime Prediction' by Hutter et al. (2014). The research demonstrates how machine learning can predict algorithm runtime on unseen inputs, e.g., estimating SAT solving time. The paper introduces various modeling techniques and conducts extensive experiments on SAT, TSP, and MIP problems, significantly improving prediction accuracy.

Original post →

More from Research

Research channel →