CIRCULATED DEEP REINFORCEMENT LEARNING UTILIZING TENSORFLOW |
Author(s): |
P.Nandhini |
Keywords: |
Deep Reinforcement Learning, Tensorflow, Deep Q-Networks, Deep Q-Learning, Artificial Generalized Intelligence |
Abstract |
Profound Reinforcement Learning is the mix of Support Learning calculations with Deep neural system, which had late achievement in learning confounded obscure conditions. The prepared model is a Convolutional Neural System prepared utilizing Q-Learning Loss esteem. The operator takes in perception, for example crude pixel picture and reward from the condition for each progression as info. The profound Q-learning calculation gives out the ideal activity for each perception and compensates pair. The hyper parameters of Deep Q-Network remain unaltered for any condition. Tensorflow, an open source AI and numerical calculation library is utilized to execute the profound Q-Learning calculation on GPU. The dispersed Tensorflow engineering is utilized to expand the equipment asset usage and lessen the preparation time. The utilization of Graphics Processing Unit (GPU) in the disseminated condition quickened the preparation of profound Q-arrange. On executing the profound Q-learning calculation for some conditions from OpenAI Gym, the operator outflanks a not too bad human reference player with few days of preparing. |
Other Details |
Paper ID: IJSARTV Published in: Volume : 5, Issue : 3 Publication Date: 3/4/2019 |
Article Preview |
Download Article |