We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.ET

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Emerging Technologies

Title: All-Spin Bayesian Neural Networks

Abstract: Probabilistic machine learning enabled by the Bayesian formulation has recently gained significant attention in the domain of automated reasoning and decision-making. While impressive strides have been made recently to scale up the performance of deep Bayesian neural networks, they have been primarily standalone software efforts without any regard to the underlying hardware implementation. In this paper, we propose an "All-Spin" Bayesian Neural Network where the underlying spintronic hardware provides a better match to the Bayesian computing models. To the best of our knowledge, this is the first exploration of a Bayesian neural hardware accelerator enabled by emerging post-CMOS technologies. We develop an experimentally calibrated device-circuit-algorithm co-simulation framework and demonstrate $24\times$ reduction in energy consumption against an iso-network CMOS baseline implementation.
Subjects: Emerging Technologies (cs.ET)
Cite as: arXiv:1911.05828 [cs.ET]
  (or arXiv:1911.05828v4 [cs.ET] for this version)

Submission history

From: Abhronil Sengupta [view email]
[v1] Wed, 13 Nov 2019 21:51:18 GMT (1539kb,D)
[v2] Thu, 21 Nov 2019 18:05:34 GMT (1539kb,D)
[v3] Thu, 2 Jan 2020 05:26:20 GMT (1728kb,D)
[v4] Fri, 17 Jan 2020 15:49:40 GMT (1728kb,D)

Link back to: arXiv, form interface, contact.