Resource-Aware Pareto-Optimal Automated Machine Learning Platform
This addresses the need for automated machine learning that balances performance with resource limitations, though it appears incremental as it builds on existing AutoML and multi-objective optimization methods.
The authors tackled the problem of building machine learning models under multiple objectives and resource constraints by introducing RA-AutoML, a platform that uses MOBOGA for hyper-parameter and neural architecture search, achieving very good accuracy on CIFAR-10 while adhering to model size constraints.
In this study, we introduce a novel platform Resource-Aware AutoML (RA-AutoML) which enables flexible and generalized algorithms to build machine learning models subjected to multiple objectives, as well as resource and hard-ware constraints. RA-AutoML intelligently conducts Hyper-Parameter Search(HPS) as well as Neural Architecture Search (NAS) to build models optimizing predefined objectives. RA-AutoML is a versatile framework that allows user to prescribe many resource/hardware constraints along with objectives demanded by the problem at hand or business requirements. At its core, RA-AutoML relies on our in-house search-engine algorithm,MOBOGA, which combines a modified constraint-aware Bayesian Optimization and Genetic Algorithm to construct Pareto optimal candidates. Our experiments on CIFAR-10 dataset shows very good accuracy compared to results obtained by state-of-art neural network models, while subjected to resource constraints in the form of model size.