Reinforcement Learning: An Introduction [Meeting 2]

Chapters 2.4 -> 3.3.

As we completed chapter 2, we learned about some improvements to the MAB problem - but also how these didn't really generalise.

Richard presented a notebook drawing the graphs from the first section:

• https://github.com/richardcooper/reinforcment-learning/blob/master/scripts/multi_armed_bandits.py
• https://sheltered-reef-51643.herokuapp.com

As we started chapter three, we covered the start of a new definition for 'agents' learning from their environment.

We made comparisons to FSMs and guessed somewhat about the formulating of rewards (a topic we expect to cover in time).

There were no photos.