Reinforcement learning (1)(2018-08-23 cont.)

Reason for the series


OK this article only serve as my study notes for reinforcement learning, there are two reasons to even let me want to write blog to learn the material. Firstly, Huge amount works show the trend of merging the three big, reinforcement, supervised and unsupervised learning. 

Second because I recently have been dealing with unsupervised learning which need some insight of how we can utilizes or construct data without the label. Since there are lots of meta-similarity going on between reinforcement learning with weakly, semi-, and self-supervised approach. Why don't we just take look at how reinforcement learning works  


#1 General Introduction


The first section will be the brief introduction on what is reinforcement learning (RL), what type are there, and lite history of Reinforcement learning. 

 * feedback interactive driven training network vs label training network

so the start point: reinforcement learning is feedback interactive driven training meaning you don't have to label the data but you have to construct the learning model or you understand what is the scenario


#2 second section 


This section we main talk about the three basic type of Reinforcement learning

a. value function 

b. policy searching 

c. AC meaning combination of the two ahead.

你可能感兴趣的:(Reinforcement learning (1)(2018-08-23 cont.))