I am working on a Artificial intillegence project and i need someone who can explain me few doubts i have.

The following paper introduced the first and excellent Deep Learning Model for Reinforcement Learning and showed to learn difficult control policies with raw pixels of input for the many Atari games. Review this excellent paper from DeepMind, “Playing Atari with Deep Reinforcement Learning”. And i have few doubts in the research paper and i need someone to explain it to me.