Transfer Learning Made Easy

Posted by Sivylla Paraskevopoulou, June 11, 2023

9 views (last 30 days) | 0 Likes | 0 comment

Transfer learning is the process of fine tuning a pretrained model to your task. In this blog post, we will show you how to modify a pretrained convolutional neural network for classifying new images.

With MATLAB, you can perform transfer learning either programmatically (at the command line) or interactively (with a low-code app). Johanna and I have prepared videos that will walk you through the steps of each workflow. In both workflows you will notice how MATLAB tools make data preparation, model surgery, and model re-training easy.

Transfer Learning at the Command Line

And now we will walk you through how to do the same workflow interactively with the Deep Network Designer app.

Transfer Learning Using an App

To replicate the workflows seen in the above videos, get the datasets and some extra help from the documentation examples Train Deep Learning Network to Classify New Images (command line) and Transfer Learning with Deep Network Designer (app).