search

mbhushan / ml

0 stars 0 forks source link

Udacity: Deep Learning: Reinforcement Learning: Monte Carlo Methods #23

Open mbhushan opened 6 years ago

mbhushan commented 6 years ago

MC control - constant aplha

screen shot 2018-05-12 at 10 47 23 am screen shot 2018-05-12 at 10 50 24 am

Epsilon Greedy Policy:

screen shot 2018-05-06 at 12 54 58 pm screen shot 2018-05-06 at 12 21 39 pm screen shot 2018-05-06 at 12 56 36 pm

Incremental Mean:

screen shot 2018-05-06 at 9 55 12 am screen shot 2018-05-06 at 9 59 57 am screen shot 2018-05-06 at 10 00 44 am

Generalized policy iteration:

screen shot 2018-05-06 at 9 47 28 am screen shot 2018-05-06 at 9 48 58 am

MC Prediction: action values:

screen shot 2018-05-06 at 9 00 07 am screen shot 2018-05-06 at 9 02 52 am screen shot 2018-05-06 at 9 01 08 am screen shot 2018-05-06 at 9 14 53 am screen shot 2018-05-06 at 9 09 49 am screen shot 2018-05-06 at 9 02 52 am

MC prediction: state values:

screen shot 2018-05-05 at 8 13 23 pm screen shot 2018-05-05 at 8 14 50 pm screen shot 2018-05-05 at 8 21 04 pm

The on and off policy methods:

screen shot 2018-05-05 at 8 07 18 pm screen shot 2018-05-05 at 8 07 08 pm screen shot 2018-05-05 at 8 06 05 pm screen shot 2018-05-05 at 8 03 34 pm