23:27 Very helpful explanation. 1 Question : Will discounting affect the behavior that we observe here? Will the agent prefer a faster, although riskier, route?
It will affect the behavior. The agent will prefer policies that lead to rewards being obtained earlier in time (they can be but are not necessarily more risky).
Absolutely marvellous introduction, Professor. Thank you so much for these insightful lectures.
23:27 Very helpful explanation. 1 Question : Will discounting affect the behavior that we observe here? Will the agent prefer a faster, although riskier, route?
It will affect the behavior. The agent will prefer policies that lead to rewards being obtained earlier in time (they can be but are not necessarily more risky).
Great introduction!
Thank you!
does mdp work for continuous states and continuous action? like work on R^2 plane instead of a finite grid
Why not just move the charging station to the upper left corner :D Great lecture btw
your boss is gonna give you really big negative reward for changing the infrastructure