Research at FAST

9 Impact Stories AMA is devoted to developing practical solutions to problems through mathematical techniques and models. We have collaborated with different industrial partners to conduct applied research such as: Prof. Yitang Zhang Prof. Chi-Wang Shu Improving Fabric Manufacture Efficiency Using a Hybrid Quality Rate Prediction Model Esquel (a world leading producer of premium cotton shirts) approached us in late 2017 requesting a data-driven process improvement that could reduce fabric wastage. Accurately predicting the A-Grade fabric wastage rate is vital in Esquel’s daily fabric manufacturing. After receiving an A-Grade fabric order, they must estimate imperfections in some of the fabric produced, and produce an additional percentage to fulfill the order. Dr Binyan Jiang and Dr Daihai He led a team to find a statistical solution to this problem. They proposed a hybrid model for predicting the A-Grade fabric rate using penalization techniques to conduct simultaneous model selection and estimation. The team developed a package delivered to Esquel at the end of June 2018 and immediately implemented in all their woven mill factories. This hybrid model and software package have helped Esquel save over HK$2.5 million annually in production costs, reduce fabric wastage by 120,000 yards per year, and produce environmental savings including a 375 ton greenhouse gas reduction, all in addition to saving manpower. Mathematical Epidemic Modelings Vaccine-preventable infectious diseases kill more than 1.5 million people worldwide annually. Understanding transmission patterns is a powerful weapon in developing effective policies to limit mortality and prevent diseases spreading. AMA research led by Dr Daihai He has mathematically modelled Chickenpox, the Zika virus and Yellow Fever transmission. We have developed new mathematical models fitted to reported epidemiological data via a likelihood-based inference framework through iterated filtering and built a tool that captures the observed epidemic and simulates different control measure outcome so the best control method can be identified and implemented. PolyU’s mathematical disease outbreak modelling has strengthened Shenzhen’s chickenpox vaccination strategy. Our model has informed the Shenzhen Centre for Disease Control and Prevention’s (SZCDC) adoption of new policies to prevent an estimated 8000 chickenpox cases annually. PolyU modelling also supported guidance on reducing Zika virus transmission including European Commission advice and provided evidence to support public health risk-mitigation and planning for Yellow Fever outbreaks in Angola and Nigeria. Moving Forward We will further: • strengthen our 3 research areas through integrating projects between them, supporting our staff to publish high quality papers, conducting collaborative research, facilitating research visits, hosting expert visitors and international conferences and further solidifying our status as a leading international research centre. • recruit young data science and analytics talent, as well as established researchers, to conduct leading research that bridges our 3 research areas and to develop leading techniques for the high-tech industry. • develop and promote interdisciplinary research through enhancing collaborations with other PolyU departments as well as other academic institutes, with special emphasis on mathematical theory, algorithms and software for handling optimization problems, statistical learning and dynamic systems in data science and artificial intelligence. • develop and create fresh links with industry and other non-academic entities to impactfully contribute to society with a special focus on collaborations in the rapidly advancing Greater Bay Area. • recruit, and provide expert training to high quality PhD students and new researchers for their future employment in top universities and leading high-tech industries. Overview