Academic lecture of College of Cyber Security

Date：November 29, 2019
Time：15:00
Venue: Room 124, Nanhai building，Main Campus

Topic 1: Granularity Fuzzy Rule Model: Design and Evaluation
Abstract: In recent years, many machine learning algorithms can use the internal representations of massive data to build the world model, leading the trend of data-driven artificial intelligence technology development and application. Granular computing is an intelligent method to simulate the fuzzy cognitive mode of human beings, which can better analyze and solve complex problems by abstracting, dividing and transforming complex problems into semantic description forms at higher levels. The intelligent system constructed under this framework, whether it is an independent rule or overall structure of the model, reflects the human-centered design idea. This report will briefly introduce that the construction of this kind of artificial intelligence system can fully show and simulate the outstanding rational reasoning ability of human beings. Through the local interpretability, that is, the inference on a single sample and the understanding of learning, the whole learning algorithm can be globally interpretable.
Speaker: National University of Defense Technology, Hu Xingchen, Lecturer
About the speaker: Dr. Hu obtained his doctorate degree from the University of Alberta, Canada in 2017 and conducted research on intelligent systems and data mining covering the areas of granular computing, fuzzy systems, neural networks and population optimization. As the first author/ corresponding author, he published 12 papers in renowned International journals and conferences such as IEEE Transactions on Fuzzy Systems, Fuzzy Sets and Systems, Pattern Recognition and IEEE International Conference on Fuzzy Systems. He has completed or is participating in a number of projects related to granular computing or the construction of intelligent systems, including 8 projects of the National Natural Science Foundation of China, the Innovation Special Zone Project of the Science and Technology Commission of the Central Military Commission, National Major Engineering Projects and National Defense Pre-research Projects. 

Topic 2: Wireless Cooperative Spectrum Sensing Based on Graph Cut Algorithm
Abstract: Cognitive radio network can effectively solve the current spectrum sparsity problem by opportunistic access to the primary user spectrum. Through the collaborations between wireless users, the collaborative wireless spectrum can effectively ensure the reliability of spectrum detection. However, in the cognitive radio network with large space division, the spectrum states observed by different users may not be exactly the same, but there are complex associated relationships. Based on the probabilistic graphical model, this paper studies the problem of collaborative wireless spectrum sensing in heterogeneous spectrum scenarios, and implements a low-complexity state decision algorithm by graph cutting.
Speaker: National University of Defense Technology, Wu Keyu, Lecturer
About the speaker: Dr. Wu graduated from School of Electronic Science and Engineering, National University of Defense Technology in September 2014 as master of engineering, and obtained his doctor degree of philosophy from the School of Electrical Engineering and Computer Science, University of Alberta, Canada in September 2018. He began his work at the National University of Defense Technology since December 2018 with his research interests covering machine learning, optimal decision theory and its application in information systems. He has published 7 academic papers in leading international journals and conferences as first author/corresponding author.

Topic3：Coded Caching under Arbitrary Popularity Distributions
Speaker: Zhang Jinbei, Sun Yat-Sen University 
Abstract: Caching plays an important role in reducing the backbone traffic when serving high-volume multimedia content. Recently, a new class of coded caching schemes has received significant interest because they can exploit coded multi-cast opportunities to further reduce backbone traffic. Without considering file popularity, prior works have characterized the fundamental performance limits of coded caching through a deterministic worst-case analysis. However, when heterogeneous file popularity is taken into account, there remain open questions regarding the fundamental limits of coded caching performance. In this work, for an arbitrary popularity distribution, we first derive a new information-theoretical lower bound on the expected transmission rate of any coded caching schemes. We then show that a simple coded-caching scheme attains an expected transmission rate that is at most a constant factor away from the lower bound. Unlike other existing studies, the constant factor that we derived is independent of the popularity distribution.
About the speaker: Dr. Zhang has been an associate professor at Sun Yat-Sen university since August 2018. He received his PhD degree from Shanghai Jiao Tong University in March 2016, and then engaged in his postdoctoral research in the department of computer science in The Chinese University of Hong Kong. His research mainly focuses on the future integration of data storage and network communication and physical layer security, which is the forefront in the area of communication network. His has published some research papers in authoritative periodicals such as TIT, TON and other authoritative academic conferences like ITA, MobiHoc and ISIT. He has also won the first excellent doctoral thesis award of communication society (2016) and was selected as the ACM MobiHoc 2018 best paper candidate.