Deep reinforcement learning



Deep learning is a new research track within the field of machine learning. The main idea behind deep learning is to create architectures consisting of multiple layers of representations in order to learn high level abstractions. Examples are the deep neural network methods used in image processing. Starting from individual pixels, each successive layer of the network learns progressively more complex features until the highest layers are able to recognize objects in the image. These networks have achieved remarkable succeses in object recognition, document classification and speech recognition tasks. One experiment performed by Google, made headlines  when their network learned to recognize cats after watching youtube for a week.


Cat concept learned by deep neural network


Reinforcement learning (RL) is one of the most promising AI paradigms for the future development of autonomous robots. RL allows a robot to learn from trial-and error interactions with its environment. By observing the results of its actions, a robot can determine the optimal sequence of actions to take in order to reach some goal.Recently the combination of deep learning and reinforcement learning was proposed. This combination allows a learning agent to control a system based only on visual inputs, using a deep neural network to extract relevant features from the images.
During this thesis you will investigate the possibilities of deep reinforcement learning by implementing and empirically evaluating deep RL architectures. This will include evaluating the algorithms on the recently proposed OpenAI Gym RL benchmark.



Peter Vrancx