A Brief Review of the ChaLearn AutoML Challenge: Any-time Any-dataset Learning Without Human Intervention

摘要: The ChaLearn AutoML Challenge team conducted a large scale evaluation of fully auto- matic, black-box learning machines for feature-based classi cation and regression problems. test bed was composed 30 data sets from wide variety application domains ranged across di erent types complexity. Over six rounds, participants succeeded in delivering software capable being trained tested without human intervention. Although improvements can still be made to close the gap between human-tweaked models, this competition contributes development automated environments by challenging practitioners solve problems under speci c constraints sharing their approaches; platform will remain available post-challenge submissions at http://codalab.org/AutoML.