Reinforcement Learning

This repository includes principal RL algorithms and my RL projects.

Gym-series environments

• Gym_CartPole_v0_DDPG
• Gym_Enduro
• Gym_Box2d
• Gym_MountainCar
• Gym_Pendulum_DoubleDQN

Distance evaluation

• Reading distance from snapshots

Mono-network (NAF)

This is an idea, not work yet. Using only one network output actions with Q values, like NAF paper did.

• mono_network
• NAF theory

Target tracking with RL

The idea based on the paper: Action-Decision Networks for Visual Tracking with Deep Reinforcement Learning Electronics and Telecommunications Research Institute ( ETRI ), South Korea;
My solution is applying the rl in continuous action area to re design this experiment

• target tracking paper

Graph signal sampling with RL

Apply the DQN to do graph sampling walk

• Graph signal sampling via DQN

Searching for optimal solution

This task is working on finding right solution of a complex function. Using the advantage of reinforcement learning, agent is used to find out the solution under complicated and tedious environment instead of human.

Multi-dimensional continuous control via hierarchical reinforcement learning frame

Apply hierarchical frame in multi-actions contorlling task.