Skip to content

gianpd/NFCP_RLmodel

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

43 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

NFCP_RLmodel

A Deep-Q-Network model for blockchain environment. Agents learn to assign a behavior value to other agents looking at agents states.

Deep-Q-Learning

nodes (agents) have to select an action for each state. States are samples reporting some features about the behaviour of each node. Actions indicate how much a node has the right to participate to the final consensus. The agent can choose between 5 actions:

     participate with 100% tickets: 0;
     participate with 75% tickets: 1;
     participate with 50% tickets: 2;
     participate with 25% tickets: 3;
     participate with 0% tickets: 4

Collect states:

StateNodeA = [1000, 40, 580, ...]
StateNodeB = [1050, 10, 340, ...]
StateNodeC = [1000, 28, 100, ...]
StateNodeD = [1050, 10, 340, ...]

A Deep Neural Network is used to approximating the Q-learning function:

About

Deep-Q-Network for blockchain environment

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages