Implementing Binarized Neural Networks with Magnetoresistive RAM without Error Correction - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

Implementing Binarized Neural Networks with Magnetoresistive RAM without Error Correction

Résumé

One of the most exciting applications of Spin Torque Magnetoresistive Random Access Memory (ST-MRAM) is the in-memory implementation of deep neural networks, which could allow improving the energy efficiency of Artificial Intelligence by orders of magnitude with regards to its implementation on computers and graphics cards. In particular, ST-MRAM could be ideal for implementing Binarized Neural Networks (BNNs), a type of deep neural networks discovered in 2016, which can achieve state-of-the-art performance with a highly reduced memory footprint with regards to conventional artificial intelligence approaches. The challenge of ST-MRAM, however, is that it is prone to write errors and usually requires the use of error correction. In this work, we show that these bit errors can be tolerated by BNNs to an outstanding level, based on examples of image recognition tasks (MNIST, CIFAR-10 and ImageNet): bit error rates of ST-MRAM up to 0.1% have little impact on recognition accuracy. The requirements for ST-MRAM are therefore considerably relaxed for BNNs with regards to traditional applications. By consequence, we show that for BNNs, ST-MRAMs can be programmed with weak (low-energy) programming conditions, without error correcting codes. We show that this result can allow the use of low energy and low area ST-MRAM cells, and show that the energy savings at the system level can reach a factor two.
Fichier principal
Vignette du fichier
NANOARCH_paper_2019.pdf (1.25 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02399718 , version 1 (09-12-2019)

Identifiants

  • HAL Id : hal-02399718 , version 1

Citer

Tifenn Hirtzlin, Bogdan Penkovsky, Jacques-Olivier Klein, Nicolas Locatelli, Adrien Vincent, et al.. Implementing Binarized Neural Networks with Magnetoresistive RAM without Error Correction. 15th IEEE / ACM International Symposium on Nanoscale Architectures (NANOARCH), Jul 2019, Qingdao, China. ⟨hal-02399718⟩
75 Consultations
72 Téléchargements

Partager

Gmail Facebook X LinkedIn More