Neural Architecture Search (NAS) is increasingly popular to automatically explore the accuracy versus computational complexity trade-off of Deep Learning (DL) architectures. When targeting tiny edge devices, the main challenge for DL deployment is matching the tight memory constraints, hence most NAS algorithms consider model size as the complexity metric. Other methods reduce the energy or latency of DL models by trading off accuracy and number of inference operations. Energy and memory are rarely considered simultaneously, in particular by low-search-cost Differentiable NAS (DNAS) solutions. We overcome this limitation proposing the first DNAS that directly addresses the most realistic scenario from a designer’s perspective: the co-optimization of accuracy and energy (or latency) under a memory constraint, determined by the target HW. We do so by combining two complexity-dependent loss functions during training, with independent strength. Testing on three edge-relevant tasks from the MLPerf Tiny benchmark suite, we obtain rich Pareto sets of architectures in the energy vs. accuracy space, with memory footprints constraints spanning from 75% to 6.25% of the baseline networks. When deployed on a commercial edge device, the STM NUCLEO-H743ZI2, our networks span a range of 2.18x in energy consumption and 4.04% in accuracy for the same memory constraint, and reduce energy by up to 2.2 × with negligible accuracy drop with respect to the baseline.

Multi-Complexity-Loss DNAS for Energy-Efficient and Memory-Constrained Deep Neural Networks / Risso, Matteo; Burrello, Alessio; Benini, Luca; Macii, Enrico; Poncino, Massimo; Jahier Pagliari, Daniele. - ELETTRONICO. - (2022), pp. 1-6. (Intervento presentato al convegno ACM/IEEE International Symposium on Low Power Electronics and Design tenutosi a Boston (USA) nel August 1 - 3, 2022) [10.1145/3531437.3539720].

Multi-Complexity-Loss DNAS for Energy-Efficient and Memory-Constrained Deep Neural Networks

Risso, Matteo;Burrello, Alessio;Benini, Luca;Macii, Enrico;Poncino, Massimo;Jahier Pagliari, Daniele
2022

Abstract

Neural Architecture Search (NAS) is increasingly popular to automatically explore the accuracy versus computational complexity trade-off of Deep Learning (DL) architectures. When targeting tiny edge devices, the main challenge for DL deployment is matching the tight memory constraints, hence most NAS algorithms consider model size as the complexity metric. Other methods reduce the energy or latency of DL models by trading off accuracy and number of inference operations. Energy and memory are rarely considered simultaneously, in particular by low-search-cost Differentiable NAS (DNAS) solutions. We overcome this limitation proposing the first DNAS that directly addresses the most realistic scenario from a designer’s perspective: the co-optimization of accuracy and energy (or latency) under a memory constraint, determined by the target HW. We do so by combining two complexity-dependent loss functions during training, with independent strength. Testing on three edge-relevant tasks from the MLPerf Tiny benchmark suite, we obtain rich Pareto sets of architectures in the energy vs. accuracy space, with memory footprints constraints spanning from 75% to 6.25% of the baseline networks. When deployed on a commercial edge device, the STM NUCLEO-H743ZI2, our networks span a range of 2.18x in energy consumption and 4.04% in accuracy for the same memory constraint, and reduce energy by up to 2.2 × with negligible accuracy drop with respect to the baseline.
2022
9781450393546
File in questo prodotto:
File Dimensione Formato  
ispled.pdf

accesso riservato

Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 1 MB
Formato Adobe PDF
1 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
ispled_open.pdf

accesso aperto

Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: Pubblico - Tutti i diritti riservati
Dimensione 707.15 kB
Formato Adobe PDF
707.15 kB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2971086