DEEP Q LEARNING AND ITS VARIANTS: A CONCISE REVIEW

Deep Q learning is one of the latest and efficient learning technique which is trending in research community. It combines the features of deep learning and Q learning algorithms. The well-known Q-learning algorithm is recognized for overvalue the action values with respect of some situations. It was not popular, such magnification are usual, whether they impair performance and whether they can be halt in General. Modern DQN algorithm, which include Q-learning (with a deep neural network(DNN)), agonize from significant over valuations in certain games like the Atari 2600.Therefore we understand the logic behind the Double Q-learning algo, it can be induce to deal with vast-scale function approximation. In this paper a certain acceptance to DQN algorithm and its variants has been analyzed and implementations are performed to show that algorithm not only reduces perceive statements, but this also escort to competitively recommended execution on different names.