multi-agent deep reinforcement learning for networked system control.

No description available

N++2 is a multi-threaded neural network simulator with network sharding and BLAS support. It was developed from 2007 to 2010 in order to realize training and realtime inference of deep autoencoder networks. With its help I created the first visual DeepRL control policies.

No description available

deeprl_sumo_docker

DeepRL Car Racing Agent is a reinforcement learning project focused on training autonomous driving agents in the CarRacing environment using Deep Q-Networks (DQN) and Proximal Policy Optimization (PPO).