Deep Neural Networks

MATLAB® makes it easy to create and modify deep neural networks. The following videos outline how to use the Deep Network Designer app, a point-and-click tool that lets you interactively work with your deep neural networks. Walk through a step-by-step example for building ResNet-18, a popular pretrained model. You’ll also learn how to easily check for errors in connections and property assignments.

In the second video, you’ll see how to use the app in a transfer learning workflow. The video demonstrates how to modify the last few layers in your imported network and then check the modified architecture for errors.

Designing Networks

Interactively Build, Visualize, and Edit Deep Learning Networks Explore a step-by-step method for building a version of ResNet-18, a popular pretrained model for deep learning. This video outlines how to check for errors in connections and property assignments using a network analyzer.

Interactively Modify a Deep Learning Network for Transfer Learning Learn how to use the Deep Network Designer app in a transfer learning workflow. This video demonstrates how to use the app modify the last few layers in the imported network, instead of modifying the layers in the command line.

Experiment Management

Interactively Build Experiments to Fine-Tune and Compare Deep Learning Networks Experiment Manager is an app for training, fine-tuning, and explaining your deep learning networks under a variety of initial conditions.

How to Set Up Your Own Deep Learning Experiments The Experiment Manager app allows you to set up experiments for training, fine-tuning, and explaining your deep learning networks under a variety of initial conditions.­