Dependable Deep Learning: Towards Cost-Efficient Resilience of Deep Neural Network Accelerators against Soft Errors and Permanent Faults.
Muhammad Abdullah HanifMuhammad ShafiquePublished in: IOLTS (2020)
Keyphrases
- deep learning
- cost efficient
- neural network
- fault diagnosis
- unsupervised feature learning
- error detection
- unsupervised learning
- deep architectures
- machine learning
- mental models
- restricted boltzmann machine
- back propagation
- deep belief networks
- expert systems
- neural nets
- weakly supervised
- maximum likelihood
- text classification
- computer vision