Please use this identifier to cite or link to this item:
https://doi.org/10.21256/zhaw-23980| Publication type: | Conference paper |
| Type of review: | Peer review (abstract) |
| Title: | Dependable neural networks through redundancy, a comparison of redundant architectures |
| Authors: | Doran, Hans Dermot Ganz, David Ielpo, Gianluca Zapke, Michael |
| et. al: | No |
| DOI: | 10.21256/zhaw-23980 |
| Proceedings: | Proceedings of the Embedded World Conference 2021 |
| Conference details: | Embedded World Conference 2021, online, 1.-5. März 2021 |
| Issue Date: | Mar-2021 |
| Publisher / Ed. Institution: | WEKA |
| Other identifiers: | arXiv:2108.02565 |
| Language: | English |
| Subjects: | Functional safety; Lockstep processor; FPGA; GPU; Machine learning; Neural network; Computer architecture |
| Subject (DDC): | 006: Special computer methods |
| Abstract: | With edge-AI finding an increasing number of real-world applications, especially in industry, the question of functionally safe applications using AI has begun to be asked. In this body of work, we explore the issue of achieving dependable operation of neural networks. We discuss the issue of dependability in general implementation terms before examining lockstep solutions. We intuit that it is not necessarily a given that two similar neural networks generate results at precisely the same time and that synchronization between the platforms will be required. We perform some preliminary measurements that may support this intuition and introduce some work in implementing lockstep neural network engines. |
| URI: | https://digitalcollection.zhaw.ch/handle/11475/23980 |
| Fulltext version: | Published version |
| License (according to publishing contract): | CC BY-NC-ND 4.0: Attribution - Non commercial - No derivatives 4.0 International |
| Departement: | School of Engineering |
| Organisational Unit: | Institute of Embedded Systems (InES) |
| Appears in collections: | Publikationen School of Engineering |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 2021_Doran-etal_Dependable-neural-networks_ew2021.pdf | 303.85 kB | Adobe PDF |  View/Open |
Show full item record
Doran, H. D., Ganz, D., Ielpo, G., & Zapke, M. (2021, March). Dependable neural networks through redundancy, a comparison of redundant architectures. Proceedings of the Embedded World Conference 2021. https://doi.org/10.21256/zhaw-23980
Doran, H.D. et al. (2021) ‘Dependable neural networks through redundancy, a comparison of redundant architectures’, in Proceedings of the Embedded World Conference 2021. WEKA. Available at: https://doi.org/10.21256/zhaw-23980.
H. D. Doran, D. Ganz, G. Ielpo, and M. Zapke, “Dependable neural networks through redundancy, a comparison of redundant architectures,” in Proceedings of the Embedded World Conference 2021, Mar. 2021. doi: 10.21256/zhaw-23980.
DORAN, Hans Dermot, David GANZ, Gianluca IELPO und Michael ZAPKE, 2021. Dependable neural networks through redundancy, a comparison of redundant architectures. In: Proceedings of the Embedded World Conference 2021. Conference paper. WEKA. März 2021
Doran, Hans Dermot, David Ganz, Gianluca Ielpo, and Michael Zapke. 2021. “Dependable Neural Networks through Redundancy, a Comparison of Redundant Architectures.” Conference paper. In Proceedings of the Embedded World Conference 2021. WEKA. https://doi.org/10.21256/zhaw-23980.
Doran, Hans Dermot, et al. “Dependable Neural Networks through Redundancy, a Comparison of Redundant Architectures.” Proceedings of the Embedded World Conference 2021, WEKA, 2021, https://doi.org/10.21256/zhaw-23980.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.