Assoc. Prof. Chuan Luo, Sichuan University, China
Biography: Dr. Chuan Luo is currently an Associate Professor with the College of Computer Science, Sichuan University, Chengdu, China. He received the Ph.D. degree in Computer Science from Southwest Jiaotong University, Chengdu, China, in 2015. He was a Visiting Ph.D. Student with the University of Regina, Regina, SK, Canada, in 2014. In Feb. 2019, he was a Visiting Scholar with the Harvard University, Cambridge, MA, USA. His current research interests include granular computing, cloud computing, and incremental learning. He won the Natural Science Prize (2nd Grade), awarded by the Ministry of Education of China (2021). He is the recipient of two Best Paper Awards at the 12th International FLINS Conference on Uncertainty Modeling in Knowledge Engineering and Decision Making (FLINS’16), and the 2012 Joint Rough Set Symposium (JRS’12), a Workshop Best Paper Award at the 2019 IEEE Cyber Science and Technology Congress (CyberSciTech’19), and two Best student Paper Awards at the 2015 International Joint Conference on Rough Sets (IJCRS’15) , and the Joint Conference of 13th China Conference on Rough Sets and Soft Computing (CRSSC’13).
He has published more than 100 research papers in international conferences and journals, such as the IEEE TKDE, IEEE TPDS, IEEE TNNLS, IEEE TFS, etc. He serves as an Area Editor of International Journal of Computational Intelligence Systems, Editor of Human-Centric Intelligent Systems, CCF Cultural Ambassador, Member of Special Committee of CAAI Granular Computing and Knowledge Discovery. He was included in the 2023 Stanford's list of World's Top 2% Scientists in the latest "single-recent-year-impact" metric. In 2024, he was elected as a backup candidate for academic and technical leaders in Sichuan Province, China.
Assoc. Prof. Xu Wu, Hainan Normal University, China
Biography: Xu Wu received her Ph.D. degree in Computer Science, from the Beijing University of Technology, in 2010. She was out of post-doctoral stations of the MOEKLINNS Lab, Department of Computer Science and Technology of Xian Jiaotong University in 2016. From July 2016 to July 2017, She worked as a visiting scholar in School of Engineering and Technology, Indiana University–Purdue University Indianapolis, Indianapolis, USA. From July 2010 to September 2017 and October 2017 to July 2023, she worked as an associate professor in Xi'an University of Posts and Telecommunications and Guangxi University, respectively. Currently, she is the leader of the Cyberspace Security discipline in School of Information Science and Technology, Hainan Normal University. Her research interests include trusted computing, pervasive computing, mobile computing, and software engineering. She has published more than 60 technical papers and books/chapters in the above areas. Her research is supported supported by National Natural Science Foundation of China (Program No. 62462029 and 62062006).
Assoc. Prof. Xijun Liang, China University of Petroleum, China
Biography: Xijun Liang is an Associate Professor and Ph.D. Supervisor at the School of Science, China University of Petroleum (East China), Qingdao, and a member of the AAAI SPC. He received his Ph.D. in Operational Research and Cybernetics from Dalian University of Technology, China, in 2013. Since 2014, he has been with China University of Petroleum, where he currently holds a faculty position. His research primarily focuses on optimization algorithms in statistical machine learning and deep learning, with notable achievements in the development of robust machine learning algorithms. Dr. Liang has published over 40 papers in prestigious journals such as IEEE Transactions on Neural Networks and Learning Systems, European Journal of Operational Research, IEEE Transactions on Industrial Informatics, Data Mining and Knowledge Discovery, Engineering Applications of Artificial Intelligence, and BMC Genomics. He has led several key research projects, including three funded by national and provincial agencies, such as the National Natural Science Foundation and the Shandong Province Natural Science Foundation Youth Fund. Additionally, he oversees a subtask of the National Key Research and Development Program focused on the safety operation and maintenance of national pipeline networks.
Speech Title: Evolutionary Strategy-based Algorithms for Optimizing Deep Neural Networks
Abstract: Deep neural networks have achieved significant breakthroughs in fields like computer vision, driving increasing interest in robust optimization algorithms. While gradient-based optimization methods are known for their fast convergence, they are prone to getting stuck in local optima or saddle points. In contrast, evolutionary algorithms (EAs) evolve populations of solutions over time to explore the global optimum, but they typically require extensive computational resources due to the need for numerous fitness evaluations, especially in high-dimensional optimization tasks. This report introduces a novel optimization approach that combines the advantages of gradient-based algorithms and evolutionary strategies to enhance the training of deep neural networks. By incorporating gradient-directed updates within the evolutionary framework, we propose algorithms with convergence guarantees designed to accelerate training and improve stability. The method is applicable to a wide range of deep network architectures, including convolutional neural networks (CNNs) and generative adversarial networks (GANs). Experimental results show that the proposed algorithms significantly improve training stability, particularly when training data is scarce, and effectively mitigate the mode collapse issue in GANs.