Original title: Quantum learning and essential cognition under the traction of meta-characteristics in an open world
Authors: Jin Wang, Changlin Song
In this article, the focus is on how AI handles the unknown—an issue called the Open World problem. While AI does well with familiar stuff, it struggles when faced with new, unexplored territories. Humans learn about new things through innate understanding, like recognizing new colors based on different cues than familiar ones. Similarly, when AI encounters new objects, it gets confused because it hasn’t learned the distinctions between these new features and the ones it knows. This article proposes a solution: an open-world model that focuses on recognizing differences in features between the old and new worlds. By using a method inspired by quantum tunneling and meta-characteristics, this model impressively learns new knowledge, like identifying people in images, achieving up to 96.71% accuracy. Essentially, it suggests that AI can learn and explore new things much like humans do.
Original article: https://arxiv.org/abs/2311.13335