Restricted Boltzmann machines can be used to carry out collaborative filtering when making recommendations. We will be using these RBMs to recommend movies to users. They are trained using ratings provided by the different users for different movies. A user would not have watched or rated all the movies, so this trained model can be used to recommend unseen movies to a user.
One of the first questions we should have is how to handle ranks in RBMs, since ranks are ordinal in nature, whereas RBMs work on binary data. The ranks can be treated as binary data, with the number of units to represent a rank being equal to the number of unique values for each rank. For example: in a rating system, where the ranks ...
