Deep Q-Learning for Home Automation
Vignesh Gokul, Parinitha Kannan, Sharath Kumar, Shomona Gracia · International Journal of Computer Applications · 2016
In this paper, the first deep reinforcement learning model for home automation systems is presented.Home automation has been one of the most important applications in the field of Artificial Intelligence.The system should learn the pattern and behaviour of the user automatically from experience and take future actions accordingly.The system proposed here makes use only of images to learn the user's needs using Deep Q-Learning, thus minimizing the use of any sensors and other hardware.The model makes use of a Convolutional Neural Network that takes as input, the image and outputs the future reward for each action.The system was tested with images of a house and describes the methods and results in the paper.