In this chapter, we saw how to develop a demo GridWorld game using DL4J, RL4J, and neural Q-learning, which acts as the Q-function. We also provided some basic theoretical background necessary for developing a deep QLearning network for playing the GridWorld game. However, we did not develop any module for visualizing the moves of the agent for the entire episodes.
In the next chapter, we will develop a very common end-to-end movie recommendation system project, but with the neural Factorization Machine (FM) algorithm. The MovieLens 1 million dataset will be used for this project. We will be using RankSys and Java-based FM libraries for predicting both movie ratings and rankings from the users. Nevertheless, Spark ML will be used for exploratory analysis of the dataset.
