Mentor: Ishika Singh
Project Members: Aaryansh Mohan Bansal
Reinforcement Learning(RL) is part of Machine Learning.RL provides very innovative algorithms for control and prediction problems.the principle of RL methods is that there is no supervisor or explicit teacher to command the correct actions.The agent learns by interacting with the environment,rewarding itself when goals are achieved and punishing itself when not. This forms the very true natural way of learning. However,over the time many algorithms have been developed in RL and its difficult to say which one is the most supreme one.The evolution of algorithms has taken place to best suit the type of problem at hand.So it becomes important for one to know which algorithm performs better in different situations.
This project aims for the same thing.I studied the different types of RL algorithms through well acknowledged RL literatures.To compare these algorithms models, many environments were created and then the algos were applied. All of the results and the environments have been well documented in the github repository of the project.