Affiliation:
1. University of Wisconsin–Madison
2. University of Wisconsin Madison
3. University of Wisconsin-Madison
Abstract
Abstract
A living animal exhibits remarkable ability to survive. It processes sensory input and takes actions to maximize the likelihood of survival. Researchers have been inspired to develop similar artificial agents powered by reinforcement learning—for instance, the Deep-Q learning agent, which learns to play Atari arcade games. In the recent development, the ability to process high-dimensional raw sensory data such as images, instead of handcrafted features, is one of the most important enablers, making it possible to train agents for different applications at scale. However, these agents are still different from fully autonomous agents such as living beings who not only process raw sensory data but also develop sensory function as part of their learning process. In this article, we show that an artificial agent powered by reinforcement learning can also spontaneously develop sensory apparatus. It can build its own bridge to connect the digital world to the physical one. This capability could be used to develop resilient agents that are adaptive in changing environments.
Publisher
Research Square Platform LLC
Reference41 articles.
1. Mastering the game of Go with deep neural networks and tree search;Silver D;Nature,2016
2. Berner, C. et al. Dota 2 with large scale deep reinforcement learning. ArXiv Prepr. ArXiv191206680 (2019).
3. Vinyals, O. et al. AlphaStar: Mastering the Real-Time Strategy Game StarCraft II. (2019).
4. Applications of deep reinforcement learning in communications and networking: A survey;Luong NC;IEEE Commun. Surv. Tutor.,2019
5. Deep reinforcement learning with its application for lung cancer detection in medical Internet of Things;Liu Z;Future Gener. Comput. Syst.,2019