Resource-Aware Pareto-Optimal Automated Machine Learning Platform

TL;DR

This paper presents RA-AutoML, a resource-aware AutoML platform that optimizes machine learning models considering multiple objectives and hardware constraints using a novel search algorithm.

Contribution

Introduction of RA-AutoML, a flexible AutoML framework that incorporates resource constraints and a new search engine MOBOGA for Pareto-optimal model generation.

Findings

Achieves high accuracy on CIFAR-10 under resource constraints

Demonstrates effectiveness of MOBOGA in multi-objective optimization

Outperforms some state-of-the-art models with resource limitations

Abstract

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.