Deep learning for content-based filtering

For the course video “Deep learning for content-based filtering”, why this is called deep learning? From what I see, it’s just very basic neural network.

Hi @Liao

It is called Deep learning because the network can be wide and deep with many layers and many units within a layers; and the features are learned to project into user and item embeddings vectors. It is a combination of two networks, user network and item network, to form a single architecture. The beauty of this is that by using network of basic form, you can create complex architecture to solve complex problems. This is the hall mark of Prof. Ng’s teaching, he always try to start with the simple, easy to understand way to introduce the concept and gradually add on to bring in the extend and depth of the subject. This MLS specialization is evidence of his way of teaching.