Paper Title
DQN-VIZ: A Deep Q-Network Visualization System

Abstract
DQN and its variations are one of the primary deep reinforcement learning algorithms used for discrete action spaces. Applying these algorithms to a particular task can be difficult and considering the number of parts involved, debugging such implementations require lot of time and effort. Our objective here is to develop a library that in addition to providing implementations of several popular variations of DQN algorithms gives access to a support system that can aid in analyzing, recording and debugging whilst applying deep reinforcement learning to the problem at hand. Keywords - Deep Reinforcement Learning, DQN, Double DQN, Dueling DQN, Recording

Author - Mohsin Wani, Mudasir Mohd., Hilal Ahmad Khanday

Published : Volume-10,Issue-10  ( Oct, 2023 )

	|		PDF	|	Viewed - 10	|	Published on 2023-12-11