Convolutional Neural Networks (CNNs) are currently one of the most widely used predictive models in machine learning. Recent studies have demonstrated that hardware faults induced by radiation fields, including cosmic rays, may significantly impact the CNN inference leading to wrong predictions. Therefore, ensuring the reliability of CNNs is crucial, especially for safety-critical systems. In the literature, several works propose reliability assessments of CNNs mainly based on statistically injected faults. This work presents a software emulator capable of injecting real faults retrieved from radiation tests. Specifically, from the device characterisation of a DRAM memory, we extracted event rates and fault models. The software emulator can reproduce their incidence and access their effect on CNN applications with a reliability assessment precision close to the physical one. Radiation-based physical injections and emulator-based injections are performed on three CNNs (LeNet-5) exploiting different data representations. Their outcomes are compared, and the software results evidence that the emulator is able to reproduce the faulty behaviours observed during the radiation tests for the targeted CNNs. This approach leads to a more concise use of radiation experiments since the extracted fault models can be reused to explore different scenarios (e.g., impact on a different application).

Emulating the Effects of Radiation-Induced Soft-Errors for the Reliability Assessment of Neural Networks / Matana Luza, Lucas; Ruospo, Annachiara; Soderstrom, Daniel; Cazzaniga, Carlo; Kastriotou, Maria; Sanchez, Ernesto; Bosio, Alberto; Dilillo, Luigi. - In: IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING. - ISSN 2168-6750. - ELETTRONICO. - 10:4(2022), pp. 1867-1882. [10.1109/TETC.2021.3116999]

Emulating the Effects of Radiation-Induced Soft-Errors for the Reliability Assessment of Neural Networks

Ruospo, Annachiara;Sanchez, Ernesto;
2022

Abstract

Convolutional Neural Networks (CNNs) are currently one of the most widely used predictive models in machine learning. Recent studies have demonstrated that hardware faults induced by radiation fields, including cosmic rays, may significantly impact the CNN inference leading to wrong predictions. Therefore, ensuring the reliability of CNNs is crucial, especially for safety-critical systems. In the literature, several works propose reliability assessments of CNNs mainly based on statistically injected faults. This work presents a software emulator capable of injecting real faults retrieved from radiation tests. Specifically, from the device characterisation of a DRAM memory, we extracted event rates and fault models. The software emulator can reproduce their incidence and access their effect on CNN applications with a reliability assessment precision close to the physical one. Radiation-based physical injections and emulator-based injections are performed on three CNNs (LeNet-5) exploiting different data representations. Their outcomes are compared, and the software results evidence that the emulator is able to reproduce the faulty behaviours observed during the radiation tests for the targeted CNNs. This approach leads to a more concise use of radiation experiments since the extracted fault models can be reused to explore different scenarios (e.g., impact on a different application).
File in questo prodotto:
File Dimensione Formato  
2021_TETC___Emulating_the_Effects_of_Radiation_Induced_Soft_Errors_for_the_Reliability_Assessment_of_Neural_Networks.pdf

non disponibili

Descrizione: Accepted Manuscript
Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 3.81 MB
Formato Adobe PDF
3.81 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
PUBLISHED_Emulating_the_Effects_of_Radiation-Induced_Soft-Errors_for_the_Reliability_Assessment_of_Neural_Networks.pdf

non disponibili

Descrizione: Versione Online
Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 7.78 MB
Formato Adobe PDF
7.78 MB 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.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2926454