查看文章

We investigated data augmentation to train neural networks (NNs) for soft-failure cause identification, demonstrating its impacts on NN complexity. Results indicate up to 68% reduction in the computational complexity of NN for each inference.