Fig. 4: Test accuracy vs. entropy for ResNet-18 on ImageNet.
From: Exploring robust architectures for deep artificial neural networks
Test accuracy is shown on the vertical axis and entropy (H) of the underlying graph is shown on the horizontal axis. The dots represent an average value calculated over five runs. The type and severity level of noise is shown on the top of each subplot. Sub-plots also show trendlines and Pearson correlation coefficients (r) with p-value. We note significant positive correlation between graph entropy and the performance of the Deep Artificial Neural Networks for all cases.
