Practical Deep Learning Examples with matlab

Handy Tools for Deep Learning
As the examples in this ebook have shown, with MATLAB you can build deep learning mod-
els without having to be an expert—and MATLAB makes light work of the more time-consum-
ing or irksome tasks in deep learning, with tools and functions for managing and labeling 
data, monitoring training progress, and visualizing results. Here’s a quick guide to tools 
we’ve used in working through the examples. 
Tool or Function
Description
ImageDataStore
Manage large sets of training and test images for deep 
learning models. 
Create a custom read function to automate preprocessing of 
multiple images
ImageLabeler
Draw boundary boxes around objects of interest.
Quickly label images at the pixel level for semantic 
segmentation.
imageDataAugmenter
Extend the training dataset by creating more test images 
through automatic translations, rotations, and scaling of 
existing images.
heatmap
Use this simple confusion matrix to visualize the accuracy of 
each category in a trained model.
deepDreamImage / 
activations
Visualize the layers of a model, and visualize the output of 
images passed through layers.
spectrogram
When working with signal data, easily convert an audio file 
into its corresponding time-localized frequency.
Learn More
A Guide to Tools and Resources for Deep Learning in MATLAB
11/19