Biography:
Honggui Han, professor, doctoral supervisor, and dean of the School of Computer Science.He has been engaged in research on intelligent control of complex systems, and has been selected forthe National Science Fund for Distinguished Young Scholars,the National Science Fund for Excellent Young Scholars, the Young Beijing Scholar, the Young Scientist of the Chinese Automation Society, and the Outstanding Young Scientist of Beijing Universities, etc.As a result ofthe research, he has published more than 100 academic papers and written 5 books; he has obtained more than 60 authorizedChinese/American invention patents,has presided over/participated in the formulation of more than 10 national/group/local standards.Hehaswon the second prize of the National Science and Technology Progress Award, the first prize of the Ministry of Education Science and Technology Progress Award, and the first prize of the Wu Wenjun Artificial Intelligence Science and Technology Progress Award, etc. He is currently the director of the "Digital Community" Engineering Research Center of the Ministry of Education and the director of the Beijing Key Laboratory of "Computational Intelligence and Intelligent Systems".He also serves as an editorial board member of journals such as China Science: Technical Sciences,IEEE Transactions on Cybernetics, etc.
Title：Intelligent optimization control for the whole process municipal wastewater treatment
Abstract：Municipalwastewatertreatment is an effective way to protect the environment and realize water resource recycling. However, due to the multi-processes, multi-working conditions, time-varying and other characteristics ofmunicipalwastewater treatment process, optimal control based on a single scale, a single level, and a single goal cannot guaranteethe optimumof theoverall operation.Multi-objective collaborative optimization control achieves multi-objective optimization between local and global, short-term and long-term, and efficiency and safety in themunicipalwastewater treatment processby constructing performance indicators at different time scales and designing a multi-conflict objective dynamic optimization method.It solves the problem of real-time dynamic optimization setting of key variables in themunicipalwastewater treatment process, and effectively reduces the operating cost.
Biography:
Wei Shen is a professor at the Institute of Artificial Intelligence, Shanghai Jiao Tong University. He received NSFC Excellent Young Scientists Fund. He has over 100 peer-reviewed publications in computer vision and machine learning related areas, including IEEE Trans. PAMI, IJCV, IEEE Trans. Medical Imaging, NeurIPS, ICML, ICCV, CVPR, etc. He served as an Area Chair for multiple top-tier international conferences, such as ICCV, CVPR, NeurIPS and ICML. He is an Associate Editor for Pattern Recognition and SCIENCE CHINA Information Sciences. He received the MICCAI Young Scientist Award in 2023 and CSIG Young Scientist Award in 2025.
Title: 3D Interactive Environment Construction for Embodied Simulation
Abstract: Embodied intelligence models require large-scale training data. Collecting data from real-world scenarios is costly and slow. Interactive 3D simulation environments, with their advantages of scalability and controllability, serve as the core data infrastructure for the scalable training of embodied intelligence models. This talk will present how the speaker's team efficiently constructs 3D interactive environments for embodied simulation from multi-view images/videos, including works on interactive 3D segmentation, 3D affordance segmentation, and interactive environment construction, as well as extended applications in medical surgical scene simulation.
Biography:
Zhang Guoqing is a Professor and Doctoral Advisor at Nanjing University of Information Science and Technology (NUIST), a recipient of the Jiangsu Outstanding Youth Fund, and a former postdoctoral researcher at Nanyang Technological University, Singapore. His long-term research focuses on person retrieval and re-identification in complex scenarios. He has served as the principal investigator for five national and provincial/ministerial research projects, including three grants from the National Natural Science Foundation of China (NSFC) across its Youth and General Programs, as well as two provincial initiatives, such as the Jiangsu Outstanding Youth Fund. Professor Zhang has published over 100 papers in top-tier AI and Computer Vision journals and conferences, including IEEE-TIP, IEEE-TMM, and IJCAI. His accomplishments have been recognized with numerous honors, including the First Prize of the Jiangsu Higher Education Scientific Research Award, the Jiangsu Information Technology Application Society Youth Science and Technology Award, and the First Prize of the 2025 Wu Wenjun AI Natural Science Award.
Title: Visually invariant representation learning inspired by cognitive mechanisms
Abstract: To address the problem of generalization failure of visual perception in the open world due to environmental degradation, internal target variation and cross-domain distribution shift, this report is inspired by the "perceptual invariance" mechanism of biological vision and systematically builds a three-layer invariant representation learning theoretical system of "physical reconstruction-semantic decoupling-generalization modeling". The report focuses on deconstructing the three core directions of overcoming imaging degradation in complex physical environments, decoupling fine-grained semantic invariance features under target deformation, and multi-modal prior-driven cross-domain invariance learning modeling. It deeply explains the physical inverse reconstruction, intrinsic semantic stripping, and multi-modal knowledge transfer mechanisms, and looks forward to new directions in future dynamic model selection from fixed models to agent-driven models.
Biography:
Rykhard Bohush is a Professor and Head of the Department of Computer Systems and Networks at Polotsk State University, Belarus, holding a Doctor of Technical Sciences degree. He graduated from Polotsk State University in 1997, earned his Candidate of Sciences (Ph.D.) degree in Information Processing from the Institute of Engineering Cybernetics of the National Academy of Sciences of Belarus in 2002, and received his Doctor of Sciences (D.Sc.) degree in Engineering in 2022. His main research areas include computer vision, image and video processing, machine learning, intelligent systems, object detection and recognition, and smart video surveillance. His recent research focuses on deep learning‑based object detection, person re‑identification and video tracking, high‑resolution image analysis, smart parking, and video smoke detection, with over 200 academic papers published.
Title: Video Smoke Detection: From Classical Computer Vision Algorithms to Transformer-Based Neural Networks
Abstract: This presentation will systematically review the evolution of smoke detection methods, from classical deterministic algorithms and neural-network-based and hybrid approaches to Transformer-based architectures. It will introduce our team’s end-to-end two-stream architecture, SMOKE-DETR. The model extracts spatial and temporal features in parallel and incorporates a physics-inspired directional constraint based on the predominantly upward motion of smoke, enhancing the representation of smoke dynamics. By integrating visual and dynamic smoke characteristics, the proposed model effectively improves smoke detection performance while maintaining a low false alarm rate.
