EC-NAS: Energy Consumption Aware Tabular Benchmarks for Neural Architecture Search

TL;DR

EC-NAS introduces an energy consumption-aware tabular benchmark for neural architecture search, enabling the development of energy-efficient models through multi-objective optimization and surrogate modeling.

Contribution

This work presents the first energy consumption-inclusive tabular benchmark for NAS, including a surrogate model and open-source dataset to promote energy-efficient neural architecture design.

Findings

Multi-objective optimization balances energy and accuracy.

EC-NAS enables identification of energy-efficient architectures.

Open-source benchmark facilitates research in sustainable AI.

Abstract

Energy consumption from the selection, training, and deployment of deep learning models has seen a significant uptick recently. This work aims to facilitate the design of energy-efficient deep learning models that require less computational resources and prioritize environmental sustainability by focusing on the energy consumption. Neural architecture search (NAS) benefits from tabular benchmarks, which evaluate NAS strategies cost-effectively through precomputed performance statistics. We advocate for including energy efficiency as an additional performance criterion in NAS. To this end, we introduce an enhanced tabular benchmark encompassing data on energy consumption for varied architectures. The benchmark, designated as EC-NAS, has been made available in an open-source format to advance research in energy-conscious NAS. EC-NAS incorporates a surrogate model to predict energy consumption, aiding in diminishing the energy expenditure of the dataset creation. Our findings emphasize the potential of EC-NAS by leveraging multi-objective optimization algorithms, revealing a balance between energy usage and accuracy. This suggests the feasibility of identifying energy-lean architectures with little or no compromise in performance.