讲主题一: Machine Learning for Target Audience Classification
主 讲 人: Dr. Raymond Chiong, Senior Lecturer, The University of Newcastle, Australia
Abstract: The vast amount and diversity of data shared on social media can pose a challenge for any business wanting to use it for identifying potential customers. In our work, we use both unsupervised and supervised learning methods to classify the target audience of a Twitter account owner with minimal annotation efforts. We first identify topic domains automatically using data shared by followers of the account owner using Twitter Latent Dirichlet Allocation (LDA). We then train a Support Vector Machine (SVM) ensemble using data from different account owners of the various topic domains identified by Twitter LDA. Experimental results show that our methods are able to successfully identify the target audience with high accuracy. In addition, we show that using a statistical inference approach such as bootstrapping in over-sampling to construct training datasets can achieve a better classifier in the SVM ensemble. This ensemble system can take advantage of data diversity, and its ability to differentiate prospective customers from the general audience may lead to the development of an application for targeted marketing, leading to business advantage in the crowded social media space.
演讲主题二: Optimization Benchmarking: An Open Source Framework for the Traveling Salesman Problem
主 讲 人: Dr. Thomas Weise, Associate Professor, University of Science and Technology, China
Abstract: Optimization algorithms have become a standard tool in many application areas such as management, logistics, engineering, design, chemistry, and medicine. They provide close-to-optimal approximate solutions for computationally hard problems within feasible time. This field has grown and evolved for the past 50 years and has several top-level journals dedicated to it. Many optimization methods are anytime algorithms, meaning that they start with a (usually bad) guess about the solution and step-by-step improve their approximation quality. All evolutionary algorithms, all local search algorithms, all swarm intelligence methods (such as ant colony optimization), CMA-ES, branch and bound, and memetic algorithms belong into this class, to name a few. The comparison and evaluation of anytime algorithms must consider the whole runtime behavior of the algorithms in order to avoid misleading conclusions. Also, performance data has to be gathered from multiple independent runs on multiple different benchmark instances. It is easy to see that a thorough analysis and comparison of optimization algorithms is complicated and cumbersome.
We present an open source software which can do this for you. You gather the data from your experiments, the software analyzes it. Our goal is to support researchers and practitioners as much as possible by automating the evaluation of experimental results. The software does not require any programming, just your benchmarking data. It imposes no limits, neither on the type of algorithms to be compared nor on the type of problem they are benchmarked on.
主 持 人: 鲍玉昆教授
Dr. Raymond Chiong is currently a senior lecturer at Faculty of Science & Information Technology, The University of Newcastle, Australia. His research interests are computational intelligence and its applications in logistics, energy and ect., evolutionary computation and optimization, business intelligence and data mining. He has published 100+ papers in international journals and conferences, and some of them appeared in IEEE Transactions on Evolutionary Computation, Information Science, Energy, Applied Soft Computing, Neurocomputing, Electronic Commerce Research and Applications, Engineering Applications of Artificial Intelligence, European Journal of Operational Research, International Journal of Production Economics and so forth. He is a senior member of IEEE and has been serving as Editor of Engineering Applications of Artificial Intelligence (Elsevier, IF:1.962), Associate Editor of IEEE Computational Intelligence Magazine (IEEE, IF:2.7), and served as guest editor for European Journal of Operational Research and International Journal of Production Economics. He has been the Co-PC Chair of IEEE Symposium on Computational Intelligence in Production and Logistics Systems (CIPLS 2014, 2013, 2011).
Dr. Thomas Weise received his Ph.D. degree from the University of Kassel, Germany, and a Diplom-Informatiker from Chemnitz' University of Technology, Germany. In 2009, he joined the University of Science and Technology in China (USTC) as a PostDoc fellow and member of the USTC-Birmingham Joint Research Institute in Intelligent Computation and Its Applications (UBRI). Since 2011, he is Associate Professor at the same group. Dr. Weise has made significant contributions to the fields of optimization, logistic planning, and evolutionary computation, and has authored/co-authored more than 80 publications in international journals and conferences. He is also the developer of an open source optimization and machine learning benchmarking tool, which mines performance information from large data sets to support researchers and practitioners in developing efficient computational intelligence solutions.