HIGHLIGHTS
- What: It necessary to develop effective strategies that between and This work analyses three reliability assessment strategies for deep neural networks and their underlying hardware highlighting the main solutions and challenges in terms of evaluation and fault characterization The authors show that an evaluation that considers the hardware characteristics is mandatory to deploy efficient software fault tolerance for large DNNs. The authors explore the benefits of combining two or more layers of fault injection abstraction to effectively assess and harden large and complex DNNs within acceptable evaluation times. The authors show how combining two experimental evaluation methods . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.