前面弄错了,应该看2017的秋季课,结果看了春季课了。

 

 

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitationCS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation       CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation neural network control a virtual robot, by imitating human motion

 

 

  

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

Domain shift cause the failure of supervised learning in imitation learning.

 

 

 

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

human expert said "turn left!!!" (step 3)

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation 

we don't want the average of the two expected behaviors. when the actions are discrete, the model works well.

 

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation  however, this is the gaussian output of continuous actions

 

 

 

 solution:

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

add a noise input here.

the defect is implicit density model is harder to train.

recommend to look at VAE and GAN and stan??? variational gradient descent, which are three methods to train implicit density models

upside: capable to mimic any form of function

downside: much more complex to implement

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

the second net is conditionally sampling from the first net

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 It's time for case study

 

 

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

this is a human with three go-pro on his head...

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation               CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

robot: 300 bucks

game control: 100 bucks

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 c for cost

r for reward (the negative of cost)

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 you know there maybe a little bit culture differences here. so like americans like to believe life is for reward, but maybe russians behavior more pessimistically. 

HAhahahahahahaha....

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

  CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

 

 

 

 

 

CS294-112 深度强化学习 秋季学期(伯克利)NO.1 Introduction NO.2 Supervised learning and imitation

 

  reinforcement learning in CS is exactly the same as optimal control in dynamic programming