What is the ε-greedy method?

・In Japanese
＜Premise knowledge＞
・Machine learning
・python

■What is the ε-greedy method?

The ε-greedy method efficiently learns the optimum action in reinforcement learning. At the beginning of learning, do not take actions based on the learning result (Q value), take actions as randomly as possible, and check the results for the actions widely. Then, as the learning progresses, the action based on the learning result is performed. It is named greedy because it widely confirms the reaction to the action.

The algorithm for calculating ε is as follows. When ε is below the threshold, the learning result is applied.

　

■Implementation example of ε-greedy method

I will explain an example used in reinforcement learning. Click here.

List of related articles TOP MENU Machine learning ・ε-greedy method ・Reinforcement learning ・Temporal Difference learning ・Regression vs Classification vs Clustering ・Parametric vs Nonparametric model ・LSTM ・Error function ・Experience Replay