Project Description:
Predict ratings for movies using the netflix movie dataset distrubted for the Netflix Prize.

Goals:

• Use a Neural Network to predict user ratings for movies.
• Beat Netflix's advertised Cinematch rating
• Determine how well a RBM works to predict ratings for movies affected by the Napolean Dynamite Problem
• Determine how well a RBM works to predict ratings for users with few ratings.

Sources:

• Hinton, et. al on RBMs
• Hinton, et. al on Contrastive Divergence
• Netflix Prize Forums on RBMs
• Sample RBM code

Results:
A 10% increase in rating accuracy over Netflix was needed to win the prize. I managed to get a 2.9% improvement.