|
|
|
Figure 3
An overview of the proposed methodology to evaluate a deep neural model with interpretation and evaluation metrics. Visual interpretation provides useful information for experts to understand how a CNN classifies a sample into a certain class, while the evaluation metric helps to understand its quantitative performance. For illustration, the figure shows an example with a natural image, where discriminative regions in an image from the `cat' class are highlighted. |
Open
access
access
journal menu![[Figure 3]](te5112fig3.jpg)



