Invited Speaker
Yuqi Chen ShanghaiTech University, China | Biography: Yuqi Chen is an assistant Professor (tenure-track) at School of Information Science and Technology in the ShanghaiTech University. Yuqi completed his Ph.D. in 2019 at the Singapore University of Technology and Design (SUTD) under the supervision of Sun Jun. Besides, he also collaborated closely with Chris Poskitt. From 2019 to 2021, he worked as a Research Scientist in the System Analysis and Verification (SAV) group at Singapore Management University(SMU). His research interest lies in cyber-physical system security in general. Specifically, he apply techniques like testing, reverse engineering, program analysis, and formal method to secure and analyze CPSs (e.g., autonomous vehicles, industrial control systems, and robotics systems). Speech Title: Efficient Detection of Toxic Prompts in Large Language Models Abstract: Large languagemodels (LLMs) like ChatGPT and Gemini have significantly advanced natural language processing, enabling various applications such as chatbots and automated content generation. However, these models can be exploited by malicious individuals who craft toxic prompts to elicit harmful or unethical responses. These individuals often employ jailbreaking techniques to bypass safety mechanisms, highlighting the need for robust toxic prompt detection methods. Existing detection techniques, both blackbox and whitebox, face challenges related to the diversity of toxic prompts, scalability, and computational efficiency. In response, we propose ToxicDetector, a lightweight greybox method designed to efficiently detect toxic prompts in LLMs. ToxicDetector leverages LLMs to create toxic concept prompts, uses embedding vectors to form feature vectors, and employs a Multi-Layer Perceptron (MLP) classifier for prompt classification. Our evaluation on various versions of the LLama models, Gemma-2, and multiple datasets demonstrates that ToxicDetector achieves a high accuracy of 96.39% and a low false positive rate of 2.00%, outperforming state-of-the-art methods. Additionally, ToxicDetector’s processing time of 0.0780 seconds per prompt makes it highly suitable for real-time applications. ToxicDetector achieves high accuracy, efficiency, and scalability, making it a practical method for toxic prompt detection in LLMs. |
Yuchang Mo School of Mathematical Sciences, Huaqiao University, China | Biography: Yuchang Mo (Senior Member, IEEE) received the BE, MS, and PhD degrees in computer science from the Harbin Institute of Technology, Harbin, China, in 2002, 2004, and 2008, respectively. He is currently a distinguished professor with the School of Mathematical Sciences, Huaqiao University, Quanzhou, China. He was a visiting scholar with the Department of Electrical and Computer Engineering, University of Massachusetts (UMass) Dartmouth, USA from September 2012 to August 2013. He was the program chair of the 5th Conference on Dependable Computing in China. He is currently a senior expert with the Reliability Research Society of China and a senior expert with the Technical Committee on Fault Tolerant Computing of China. His interests include the reliability analysis of complex systems and networks using combinatorial models. His research has been supported by the National Science Foundation of China. Speech Title: Reliability Analysis of Complex Systems and Networks Using Combinatorial Models Abstract: Based on Shannon's decomposition theorem, decision diagrams can represent logical functions as directed acyclic graphs in a form that is both compact and canonical. Following the pioneering work of implementing binary decision diagrams for fault tree analysis in 1993, multiple forms of decision diagrams have been developed for the reliability analysis of complex systems in diverse applications such as space exploration, nuclear power, wireless sensor networks, body area networks, unmanned aerial vehicles, cloud computing, social networks, Internet of Things, etc. This talk presents a systematic review of decision diagrams, classifying and reflecting on over ten years of my research work dedicated to applying this combinatorial model to the reliability analysis of various complex systems and networks. It also discusses potential directions for further advancing decision diagram-based reliability theory and practice. |
Jinfeng Gao Zhejiang Sci-Tech University, China | Biography: Jinfeng Gao received bachelor’s degree in Automation from Hebei University of Engineering in 2000, master degree and Ph.D. degree in Control Science and Engineering from Zhejiang University of Technology and Zhejiang University in 2003 and 2008, respectively. She had been as a visiting scholar at the university of Newcastle of electrical engineering college in Australia for one year. Now she is a professor at Zhejiang Sci-Tech University. Her main research interests include Robots control , networked control and multi-agent systems. Speech Title: A novel finite-time non-singular robust control for robotic manipulators Abstract: Robotic manipulators have been extensively used in the area of industry, agriculture, and medicine. Parameter uncertainties and external complex disturbances both bring challenges in achieving finite-time high precision control of robotic manipulators. This study addresses the finite-time non-singular robust control problem of robotic manipulators with parameter variations and external complex disturbances. To enable the tracking error of robotic manipulator system with uncertainties to converge within finite time, a novel finite-time non-singular robust control (NFNRC) approach is proposed. To make the tracking error of robotic manipulator system have faster convergence rate, we design a new nonlinear term in the robust control function. With Lyapunov stability theorem, the finite-time stability of the robotic manipulator system is ensured. Performance comparisons with non-singular terminal sliding mode control (NTSM) and sliding mode control (SMC) are studied on a nonlinear robotic manipulator system. The results validate the efficacy of the designed robotic manipulator control scheme. |
