报告题目1：Harmony Search for Feature Selection
Professor Qiang Shen received a PhD in Knowledge-Based Systems and a DSc in Computational Intelligence. He holds the Established Chair of Computer Science and is a University Strategic Executive member and Director of the Institute of Mathematics, Physics and Computer Science at Aberystwyth University. He is a Fellow of the Learned Society of Wales and was a UK Research Excellence Framework (2008-2014) panel member for Computer Science and Informatics. He has been a long-serving Associate Editor or Editorial Board member of many leading international journals (e.g., IEEE Transactions on Cybernetics and IEEE Transactions on Fuzzy Systems), and has chaired and given keynotes at numerous international conferences.
Professor Shen’s current research interests include: computational intelligence, learning and reasoning under uncertainty, pattern recognition, data modelling and analysis, and their applications for intelligent decision support (e.g., space exploration, crime detection, consumer profiling, systems monitoring, and medical diagnosis). He has authored 2 research monographs and over 350 peer-reviewed papers, including an award-winning IEEE Outstanding Transactions paper. He has served as the first supervisor of more than 50 PDRAs/PhDs, including one UK Distinguished Dissertation Award winner.
Feature selection (FS) addresses the problem of selecting those system descriptors that are most predictive of a given outcome. Unlike other dimensionality reduction methods, with FS the original meaning of the features is preserved. This has found application in tasks that involve datasets containing very large numbers of features that might otherwise be impractical to process (e.g., large-scale image analysis, text processing and Web content classification). As such, many strategies have been exploited in an effort to identify more compact and better quality feature subsets.
This talk will focus on a recently proposed feature selection approach that utilises a music-inspired meta-heuristic: Harmony Search. The resultant framework is structurally simple, and is capable of identifying multiple, good quality feature subsets. Its competitive performance and stochastic behaviour has also inspired a number of extensions and theoretical applications, including the generation and reduction of feature subset-based classifier ensembles, and feature selection for dynamic data sets. The talk will conclude with an outline of opportunities for further development.
报告题目2：Dynamic Fuzzy Rule Interpolation and its Application to Intrusion Detection
报告人：Changjing Shang 副教授
Dr Changjing Shang is a Senior Research Fellow with the Department of Computer Science, in the Institute of Mathematics, Physics and Computer Science at Aberystwyth University. She received an MSc with Distinction in Artificial Intelligence from the University of Edinburgh and a PhD in Computing from Heriot-Watt University. She has served as an Associate Editor for the Journal of Human-centric Computing and Information Sciences (Springer), and a guest editor or editorial board member of several other international journals.
Dr Shang’s previous research involved stochastic process analysis, adaptive learning, systems optimization and diagnostic decision-making. Her recent and current work on information granule and link-based modelling and analysis of complex data has been adopted and further developed internationally. She has published extensively and supervised more than 10 PhD students in these areas. As a Principal Investigator, Dr Shang has held a range of research grants, including prestigious awards by the Royal Academy of Engineering.
Application of fuzzy rule interpolation (FRI) has been escalating for making intelligent systems viable in solving many challenging real-world problems. However, requirements of such systems may change over time and the supporting static rule base may not be able to provide accurate interpolation results in the long run. Dynamic fuzzy rule interpolation (D-FRI) offers one of the potential solutions for such problems. One potential application is for network security that is often one of the biggest concerns of any organization irrespective of their size and nature of business.
Intrusion detection systems (IDSs) are considered as a popular and effective security tool for generating alerts to network administrators to inform possible or existing threats. A standard IDS may not be very effective or even unsuitable for an organizational or individual’s requirements. This talk will present an application of D-FRI for building an effective IDS. In particular, it will introduce an intelligent IDS that is built upon the most popular open source IDS, Snort via integration with D-FRI. The talk will illustrate the results that the integration of D-FRI with Snort provides an additional level of intelligence in predicting possible threats. This integration also facilitates a dynamic rule base, by promoting new rules based on the current network traffic conditions, which helps Snort to reduce both false positives and false negatives.