The authors present Auto-Tuned Models for automated machine learning; they describe the purpose of their research and demonstrate the effectiveness of their system compared to human-generated solutions.
In this paper, the authors present Auto-Tuned Models, or ATM, a distributed, collaborative, scalable system for automated machine learning. Users of ATM can simply upload a dataset, choose a subset of modeling methods, and choose to use ATM's hybrid Bayesian and multi-armed bandit optimization system. The distributed system works in a load-balanced fashion to quickly deliver results in the form of ready-to-predict models, confusion matrices, cross-validation results, and training timings. By automating hyperparameter tuning and model selection, ATM returns the emphasis of the machine learning workflow to its most irreducible part: feature engineering. The authors demonstrate the usefulness of ATM on 420 datasets from OpenML and train over three million classifiers. Their initial results show ATM can beat human-generated solutions for 30 percent of the datasets, and can do so in 1/100th of the time. (Published Abstract Provided)
Downloads
Similar Publications
- The Over-citation of Daubert in Forensic Anthropology
- A State-wide Analysis of Characteristics and Predictors of Dual System Involvement Among Child Victims of Human Trafficking
- From Research to Reality: Recruiting More Women into the Policing Profession: Preliminary Results of an NIJ-funded Applied Research Project