Question on the cost function on content-based filtering by using NN

Course Q&A Machine Learning Specialization Unsupervised Learning, Recommenders, Reinforcement
1

Hi~,
i have a hard time to understand the cost function of NN for content_based filtering.

image
image865×444 74.3 KB

The Vj_u dot Vi_m is to predict the probability that yi_j is 1, which is a probability, while yi_j is the rating given by user j on item i (which is a real rating number). How come these two number to subtract to get the cost? Any thing i miss here?
Thank you in advance.

2

Hi @Lanying_Ma

At timestamp 2:40 of the lecture video Deep learning for content-based filtering, prof. Ng did mention that the algorithm is modified slightly, from previously 0 to 5 stars rating to a binary label, 0 or 1. So y^{(i,j)} will have a value of 1 if the item i has been rated by user j.