Framelets on Graphs for Deep Learning Applications
DescriptionArtificial Intelligence known as AI enters the horizon of every people's eyes never as frequently as in today's modern society. The power of AI in terms of machine and deep learning reaches almost every branch of people's daily life. For example, smartphones' personal assistants based on natural language processing, autopilot cars based on object and pattern recognition, entertainment or e-shopping experiences based on recommender systems, and so on. Many concepts or inventions that only happened in scientific movies or novels before are now commonly appeared and used across the whole world. In short, "the future is now'' is indeed happening.Behind the great success of AI is of course the rapid advancement of computer hardware as well as software achievements that are driven by scientific research. The deep learning techniques together with data science are nowadays two of the powerful "engines'' for the AI development. Since data are not only defined on regular Euclidean domains but also on irregular domains such graphs, in this project, we focus on analysis and processing of data defined on graphs using framelets and consider its applications in deep learning. Framelets specifically designed for graphs will provide new and efficient graph data processing based on deep convolution neural networks (CNNs). The results of this project will contribute to the further development of AI.
|Effective start/end date||1/01/20 → …|