Random Forests (RFs) are popular Machine Learning models for edge computing, due to their lightweight nature and high accuracy on several common tasks. Large RFs however, still have significant energy costs, a serious concern for battery-operated ultra-low-power devices. Following the adaptive (or dynamic) inference paradigm, we introduce a hardware-friendly early stopping policy for RF-based classifiers, halting the execution as soon as a sufficient prediction confidence is achieved. We benchmark our approach on three state-of-the-art datasets relative to different embedded classification tasks, and deploy our models on a single core RISC-V microcontroller. We achieve an energy reduction ranging from 18% to more than 91%, with an accuracy drop lower than 0.5%. Additionally, we compare our approach with other early-stopping policies, showing that we outperform them.
Low-Overhead Early-Stopping Policies for Efficient Random Forests Inference on Microcontrollers / Daghero, F; Burrello, A; Xie, C; Benini, L; Calimera, A; Macii, E; Poncino, M; Jahier Pagliari, D (IFIP ADVANCES IN INFORMATION AND COMMUNICATION TECHNOLOGY). - In: VLSI-SoC: Technology Advancement on SoC Design / Grimblatt V., Chang C.H:, Reis R., Chattopadhyay A., Calimera A.. - STAMPA. - Cham (Switzerland) : SPRINGER INTERNATIONAL PUBLISHING AG, 2022. - ISBN 978-3-031-16817-8. - pp. 25-47 [10.1007/978-3-031-16818-5_2]
Low-Overhead Early-Stopping Policies for Efficient Random Forests Inference on Microcontrollers
Daghero, F;Burrello, A;Xie, C;Benini, L;Calimera, A;Macii, E;Poncino, M;Jahier Pagliari, D
2022
Abstract
Random Forests (RFs) are popular Machine Learning models for edge computing, due to their lightweight nature and high accuracy on several common tasks. Large RFs however, still have significant energy costs, a serious concern for battery-operated ultra-low-power devices. Following the adaptive (or dynamic) inference paradigm, we introduce a hardware-friendly early stopping policy for RF-based classifiers, halting the execution as soon as a sufficient prediction confidence is achieved. We benchmark our approach on three state-of-the-art datasets relative to different embedded classification tasks, and deploy our models on a single core RISC-V microcontroller. We achieve an energy reduction ranging from 18% to more than 91%, with an accuracy drop lower than 0.5%. Additionally, we compare our approach with other early-stopping policies, showing that we outperform them.File | Dimensione | Formato | |
---|---|---|---|
postprint.pdf
Open Access dal 23/09/2023
Descrizione: Post-Print
Tipologia:
2. Post-print / Author's Accepted Manuscript
Licenza:
PUBBLICO - Tutti i diritti riservati
Dimensione
2.8 MB
Formato
Adobe PDF
|
2.8 MB | Adobe PDF | Visualizza/Apri |
vlsi-soc-chapter.pdf
non disponibili
Descrizione: Pdf editoriale
Tipologia:
2a Post-print versione editoriale / Version of Record
Licenza:
Non Pubblico - Accesso privato/ristretto
Dimensione
738.19 kB
Formato
Adobe PDF
|
738.19 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/11583/2974818