Exploring Winograd Convolution for Cost-effective Neural Network Fault Tolerance

Open in new window