Assembled-OpenML: Creating Efficient Benchmarks for Ensembles in AutoML with OpenML
This tool addresses the problem of inefficient benchmarking for ensemble methods in AutoML, making comparisons faster and more accessible for developers, though it is incremental as it builds on existing OpenML infrastructure.
The paper tackles the computational expense of comparing ensemble techniques in AutoML by introducing Assembled-OpenML, a Python tool that builds meta-datasets from OpenML to reuse prediction data, reducing the time to gather predictions for 1523 base models on 31 datasets to about 1 hour compared to 37 minutes for training one model on the most expensive dataset.
Automated Machine Learning (AutoML) frameworks regularly use ensembles. Developers need to compare different ensemble techniques to select appropriate techniques for an AutoML framework from the many potential techniques. So far, the comparison of ensemble techniques is often computationally expensive, because many base models must be trained and evaluated one or multiple times. Therefore, we present Assembled-OpenML. Assembled-OpenML is a Python tool, which builds meta-datasets for ensembles using OpenML. A meta-dataset, called Metatask, consists of the data of an OpenML task, the task's dataset, and prediction data from model evaluations for the task. We can make the comparison of ensemble techniques computationally cheaper by using the predictions stored in a metatask instead of training and evaluating base models. To introduce Assembled-OpenML, we describe the first version of our tool. Moreover, we present an example of using Assembled-OpenML to compare a set of ensemble techniques. For this example comparison, we built a benchmark using Assembled-OpenML and implemented ensemble techniques expecting predictions instead of base models as input. In our example comparison, we gathered the prediction data of $1523$ base models for $31$ datasets. Obtaining the prediction data for all base models using Assembled-OpenML took ${\sim} 1$ hour in total. In comparison, obtaining the prediction data by training and evaluating just one base model on the most computationally expensive dataset took ${\sim} 37$ minutes.