What is Transfer Learning in Machine Learning?

While Deep Learning has been getting all the limelight lately in the AI field, there is another method that has famously been named by Andrew Ng, 'the next driver of Machine Learning's success'.

The easiest way of explaining Transfer Learning is with an example. Imagine when you first learned to ride a bike - you had to learn everything from scratch.

Fast forward 20 years and now you want to learn how to ride a motorbike. You don't need to start from zero. You already have knowledge of how to keep balance, use the brakes and steer the vehicle.

You can transfer the skills you learned before, doing a different but related activity.

Transfer Learning is therefore the improvement of learning in a new task through the transfer of knowledge from a related task that has already been learned.

Yes, machines can do this, too!

In this video, we show how traditional ML works and how this contrasts with the new world of Transfer Learning.

In a nutshell, using knowledge already gleaned from the successful execution of task 1, an algorithm can even require smaller of a dataset for the successful execution of learning system task 2.

The model (while not the AI itself) is general instead of specific.

Pretty neat, if you ask us.

Let us know if you have any questions after watching this informative, introductory video on a really interesting ML topic.

#transferlearning #machinelearning #AI

___________________________________________

Levity is an AI software that allows you to create custom models to train for classification of image, documents and text.

Levity is completely no-code, meaning you don’t need to have any prior knowledge of machine learning or coding algorithms!

Our tool is intended for non-technical process managers who are part of small- and mid-size organizations and banish repetitive and mind-numbing tasks from their workdays entirely.

Give us a follow on socials! We're active almost everyday and love chatting to our community members!