Visualization and Verification

Visualize neural network behavior, explain predictions, and verify robustness using image data

Visualize deep networks during and after training. Monitor training progress using built-in plots of network accuracy and loss. To investigate trained networks, you can use visualization techniques such as Grad-CAM, occlusion sensitivity, LIME, and deep dream.

Use deep learning verification methods to assess the properties of deep neural networks. For example, you can verify the robustness properties of a network, compute network output bounds, and find adversarial examples.

Featured Examples

Grad-CAM Reveals the Why Behind Deep Learning Decisions

Use the gradient-weighted class activation mapping (Grad-CAM) technique to understand why a deep learning network makes its classification decisions. Grad-CAM, invented by Selvaraju and coauthors [1], uses the gradient of the classification score with respect to the convolutional features determined by the network in order to understand which parts of the image are most important for classification. This example uses the GoogLeNet pretrained network for images.