Biograph: Prof. Caiming Qiu, IEEE fellow, Distinguished Professor of National Thousand Talents Program, Director of Research Center for Big Data of Shanghai Jiaotong University. He contributed to the development of the rigorous analysis of big data under random matrix framework. All of the results are included in 3 monographs titled as Smart Grid and Big Data: Theory and Practice, Cognitive Networked Sensing and Big Data and Cognitive Radio Communication and Networking: Principles and Practice. His current interest is in some theoretical problems arising in deep learning and big data. Prof. Caiming Qiu holds 8 patents and published over 70 journal papers and 100 conference papers. He won ICC Best Paper Award in 2011. ICC is one of the IEEE Communications Society’s two flagship conferences.
Prof. Caiming Qiu received the Ph.D. degree in electrical engineering from New York University. He served at Bell Laboratories from 1997 to 2000. He was Founder-CEO and President of Wiscom Technologies, Inc.
Title: Toward theoretical understanding of deep learning
Abstract: Nowadays, deep learning is definitely one of the most popular topics in computer science. For some specific tasks, e.g., speech recognition, image classification, deep learning is the state-of-the-art approach, even better than human beings. However, there is still no significant result on the foundation of deep learning. Recently, more and more works focus on the theoretical analysis and interpretation of deep learning from various viewpoints.
In this talk, we will first give a brief review on the theoretical development of deep learning. Several emerging approaches or frameworks, e.g., RMT(Random Matrix Theory), NTK(Neural Tangent Kernel) are introduced. Finally, we will demo some interested applications of deep learning in image understanding, anomaly detection and wireless communication.