Hi, i have a question regarding collaborative algorithm.

Here, feature vector (*x*) and user vector (*w* and *b* params) are learned simultaneously from rating data. All these vectors are initialized with random values.

So… how can we make sure after training *x* represents feature vector of movie and *w* and *b* are for the users.

You know this because of the sizes of the X and W matrices. One of them uses the number of users, and the other uses the number of movies.

Thanks for the quick reply.

Okay, I think i understood the logic, but what if both are same… say a 100 users and 100 movies

That’s just a happy coincidence. It doesn’t change which data is in which rows or columns.

Okay, thank you.