Deep Reinforcement Learning
As part of Udacity's Deep Reinforcement Learning nanodegree program, I created three projects trained to perform tasks in Unity's learning environments. Below is a description of the three projects, as well as images captured from my trained agents.
The first project is an implementation of deep Q-learning, trained on the Banana Collector environment. In this environment the agent is a robot attempting to collect yellow bananas while avoiding the blue bananas.
The code and more information can be found here.
The second project is an implementation of PPO with an actor-critic network. It is trained on the Reacher environment, in which each arm is an agent with the goal of keeping the end of the arm inside the moving sphere.
The code and more information can be found here.
The third project is an implementation of DDPG. This is a multi-agent environment in which each agent is rewarded for hitting the ball over the net.
The code and more information can be found here.
