In this unit we introduce one of the most powerful concepts in machine learning: the embedding. It’s the idea that instead of being explicitly given a representation for something, we learn it from data based on the properties it should have. We’ve actually already seen embeddings of various kinds (the inputs to the last layer of an image classifier for one), but we’ll look at two examples where the embedding aspect is even clearer: movies and words.
Students who complete this unit will demonstrate that they can:
Describe how embeddings are used in recommender systems.
Describe how embeddings can be learned from data by gradient descent.
Explain both intuitively and mathematically what it means for the vectors for two items to be “close”.
Explain how a trained decision tree makes a prediction.
Fit and evaluate a model on tabular data using Scikit-Learn
Their DotProduct class doesn’t implement a dot product; it uses it for a rating prediction task.
Think about how this task is almost equivalent to a logistic regression that takes a user embedding and predicts the score of each of the movie classes: the shape of the movie weights and biases are the same. What’s different?