3 years ago

Category: Software, Data, machine-learning, artificial-intelligence

Tales of an exploration on antigravity and other potentially unrelated matters Deep Reinforcement Learning is a somewhat new field within Machine Learning or Artificial Intelligence (you may pick your favorite term between these two, even if they’re not strictly the same), which combines Deep Learning and Reinforcement Learning and is based on the general idea that an agent can learn by observing its actions and their consequences.

This is not, in my opinion, a book intended for fast reading, and it’s ok to go back and read again some of the previous pages to let this new terminology slowly take shape in your mind.

That allows to create a reproducible environment to play around with implementations of some of the algorithms described.

The four chapters in last part of the book (“Environment Design”) return to some of the concepts that were presented earlier (states, actions, rewards and transition function) and give some practical examples.