Data Aware Neural Architecture Search

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

Abstract

Neural Architecture Search (NAS) is a popular tool for automatically generating Neural Network (NN) architectures. In early NAS works, these tools typically optimized NN architectures for a single metric, such as accuracy. However, in the case of resource constrained Machine Learning, one single metric is not enough to evaluate a NN architecture. For example, a NN model achieving a high accuracy is not useful if it does not fit inside the flash memory of a given system. Therefore, recent works on NAS for resource constrained systems have investigated various approaches to optimize for multiple metrics. In this paper, we propose that, on top of these approaches, it could be beneficial for NAS optimization of resource constrained systems to also consider input data granularity. We name such a system “Data Aware NAS”, and we provide experimental evidence of its benefits by comparing it to traditional NAS.
Original languageEnglish
Title of host publicationProceedings of tinyML Research Symposium
Number of pages7
Publication statusAccepted/In press - 2023
EventtinyML Research Symposium’23 - Burlingame, United States
Duration: 27 Mar 202327 Mar 2023

Conference

ConferencetinyML Research Symposium’23
Country/TerritoryUnited States
CityBurlingame
Period27/03/202327/03/2023

Keywords

  • Neural Architecture Search
  • Resource Constrained Machine Learning
  • tinyML
  • Convolutional Neural Networks
  • Data Granularity

Fingerprint

Dive into the research topics of 'Data Aware Neural Architecture Search'. Together they form a unique fingerprint.

Cite this