Learning with small samples

Machine Learning with small samples, including zero-shot learning


Build a powerful image classifier, using only small amount of training examples.

Use of data augmentation to eliminate model overfitting.

Leverage various auxiliary sources of information to tackle zero-shot learning problem.

Hands-on experience to train text embedding models using python.

How to use a pre-trained text embedding in a neural network.


The ML with small samples introduces you to zero-shot and few-shots current state of the art zero-shot and few-shot learning models. 

You will become familiar with different techniques on how to overcome the limitation of small sized datasets. 

This course also gives you an overview about text embedding models, how to train them from large text corpus, and how to use pre-trained ones to solve various problems including zero-shot image classification.