Prof. Qiao Chen

Xi'an Jiaotong University, China

Qiao Chen is a full professor of the School of Mathematics and Statistics, Xi'an Jiaotong University. She achieved her bachelor's, master's, and doctoral degrees all from Xi'an Jiaotong University. She conducted her postdoctoral research in the department of Biomedical Engineering at Tulane University in the United States. As research fellows, she has visited and collaborated with institutions of the University of Liverpool in the United Kingdom, the University of Queensland and the University of Technology Sydney in Australia, and Nanyang Technological University in Singapore. Her current research focuses on the mathematical foundations of artificial intelligence, brain-inspired intelligence, and neuroimaging. She has published over 40 academic papers on deep learning, neural networks and brain science in many prominent international journals, such as IEEE Trans. Biomedical Engineering, Medical Image Analysis, Neural Networks, IEEE Signal Processing Magazine. Her research is founded by the National Natural Science Foundation of China, the National Key Research and Development Program of China, the Natural Science Basis Research Plan in Shaanxi Province of China, and several industry-sponsored projects.

Speech Title: " Spatio-Temporal Deep Learning with Explainability"

Abstract: Spatio-temporal data carry a wealth of inherent spatio-temporal coupling relationships, and its modeling via deep learning is essential for uncovering the systems’ spatio-temporal evolution mechanisms. However, current models for mining spatio-temporal information usually overlook their intrinsic coupling associations, providing an incomplete understanding of the coupling mechanisms. Meanwhile, research on the explainability of spatio-temporal learning models is limited. In this talk, I will present an explainable spatio-temporal coupling learning framework. This framework constructs a deeplearning system based on spatio-temporal correlation that enables the detection of dynamic interactions and updates the spatio-temporal information to enhance the learning of their coupling relationships. Furthermore, this framework explores spatio-temporal evolution at each time step, providing a better explainability of the learning results. Finally, the proposed framework is applied to brain dynamic functional connectivity (FC) analysis. Experimental results demonstrate that the proposed model can effectively capture the variations in dynamic FC during development and the evolution of spatio-temporal information of the brain during the resting state. Two distinct developmental functional connectivity patterns are shown. Specifically, the connectivity among regions related to emotional regulation decreases, while the connectivity among regions associated with cognitive activities increases.

 

Assoc. Prof. Shinji Kawakura

The Kyoto College of Graduate Studies for Informatics, Japan

Shinji Kawakura is Associate Professor at The Kyoto College of Graduate Studies for Informatics, Kyoto-Fu, Kyoto, Japan. He got his Ph.D. in Environmentology, University of Tokyo, 2015, Bunkyo-ku, Tokyo, Japan, B.A. in Control System Engineering, Tokyo Institute of Technology, 2003, Meguro-ku, Tokyo, Japan, M.A. in Human-Factor Engineering, Tokyo Institute of Technology, 2005, Meguro-ku, Tokyo, Japan. His research areas are Informatics, systems engineering, in particular various sensing systems. He has researched in diverse International and Japanese research projects. He is the author of 31 academic papers. He has published 3 patents in Japan. He is a senior member of IEEE and Hong Kong Chemical, Biological & Environmental Engineering Society (HKCBEES).

Speech title: "Data-acquisition and Data-Processing Methodologies Utilizing Machine Learning-Based, Particularly Deep Learning-Based Techniques in the Field of Various Human Sensing"

Abstract: In recent years, as for our research, we have focused on developing application methodologies for advancements and security issues related to agricultural fields, nursing-care facilities, hospitals, for physically challenged people, and human dynamics fields. Particularly, we have focused on aquaphotomics fields, for instance, on medical fields, food fields and cosmetics, mainly skin care fields, etc. using diverse analyses methods. For the purpose, we have developed and used diverse systems: (1) directly attached sensor-based methodologies and systems like wearable sensing systems including contact sensors, (2) indirect sensor-based methodologies and systems (e.g., both specific and non-specific digital camera, infrared sensors-based systems), we have embedded them onto edge-computing systems and (3) methodologies related fields that performs various moisture (water) analyses using near infrared rays (NIR) and other methods. As for the targets of measurement, we selected (1) real agricultural workers bodies, (2) their diverse tools and utilities, (3) general harvests and the surrounding things, (4) a human-shaped specific rubber model, (5) walking support systems and (6) real human body, particularly, water in a human body and other animals. Furthermore, we have analyzed these datasets using machine learning (deep learning)-based methods (CNN, RNN, LSTMs, GANs, Explainable Artificial Intelligences (XAIs), etc.), and have provided meaningful suggestions and cues output from the analyses for managers and other members working for aforementioned places.

 

Dr. Claudia Cianci

Polytechnic University of Bari, Italy

Claudia Cianci is a Research Fellow at the Department of Mechanics, Mathematics, and Management (DMMM) of the Polytechnic University of Bari. She obtained her bachelor's, master's and doctorate degrees, all from the Polytechnic of Bari. She has actively collaborated with the University of Bari and the University of Foggia. Currently, she is engaged in postdoctoral research at the DMMM. Her research is centered around two primary areas: the mechanical performance of clear dental aligners and the mechanical behavior of cancer cells subjected to ultrasonic waves. Claudia has contributed extensively to these fields, publishing over 7 academic papers in renowned international journals, including the Journal of the Mechanical Behavior of Biomedical Materials, Journal of Cancer, Nanomaterials, and Frontiers in Materials.

Speech Title: "Damage Source Localization by Acoustic Signal Processing: an Experimental Study on Clear Dental Aligners"

Abstract: Clear aligners have gained popularity in recent years as an aesthetically pleasing option for orthodontic treatment. However, their efficacy in treating dental malposition and malocclusion has yet to be extensively studied. Clear aligners are primarily made from thermoplastic materials that provide adequate elasticity while minimizing plastic deformation during use. This design ensures a stable force is applied to the teeth, facilitating their repositioning. Our study aims to investigate the mechanical behaviour of these devices over time by comparing aligners made from two different thermoplastic materials: polyethylene terephthalate-glycol modified (PET-G) from Dentsply Sirona and polyurethane (PU) from Ghost Aligners. The aligners were subjected to cyclic compression loading to simulate the swallowing process over 15 days of use, surrounded by artificial saliva to replicate the intraoral environment. We employed the Acoustic Emission (AE) technique to analyse damage progression in the aligners during loading. The AE results were further compared with the energy absorbed and changes in stiffness of the aligners. Moreover, we validated the evolution of damage post-loading using optical microscopy. Our findings indicate a strong correlation between the AE results and the mechanical and optical microscopy data, enhancing the understanding of the mechanical behaviour of clear aligners.