Decision science is concerning with the scientific methods of decision making, which includes decisions under uncertainties, information and decision support systems, theoretical work in Markov decision processes, stochastic programming, and data science. The goal is to develop theories and start-of-the-art methods to extract knowledge from data and use that knowledge for robust manufacturing and business decisions.
In NTUIE, decision science research concentrates on the following research fields:
- Statistics and machine learning methods for knowledge extraction: Efficient knowledge extraction is the fundamental building block for intelligent manufacturing/business decisions. The lack of general-purpose knowledge extraction algorithms means that the use of historical structured/unstructured data, which widely exists in manufacturing and business environments, has been challenging in the past. Thus, the recent research interests are to develop efficient prediction engines that can extract knowledge from structured/unstructured data. The algorithms and methods. Those methods have been transferred to industry partners from the semiconductor, steel processing, textile, and retail industries.
- Stochastic optimization methods for robust manufacturing and business decisions: Knowledge extracted from historical data is often an inaccurate estimation of system characteristics, so the use of stochastic and robust optimization methods is the key to the successful implementation of decision support systems in manufacturing and business systems. Moreover, internal or external uncertainties, such as uncertain demand or machine reliability, also hamper the performance of manufacturing/service systems substantially. Thus, robust decisions are crucial for the long-term success of any business in those stochastic environments. The algorithms and methods developed in NTU have been endorsed by industry partners from the telecommunication, electronic manufacturing service (EMS), semiconductor, and automobile industries.
The consistent track record of NTUIE in decision science research has led to several industry-funded projects. Through industry collaboration, the solutions and research results to which he contributed have been successfully adopted in industrial applications through research projects funded by world-renowned companies:
- Supply chain management for electronics and automobile industries
- CISCO Systems INC.
- Delta Electronics
- Daimler Mobility AG. (a subsidiary of Mercedes-Benz / Daimler AG)
- Intelligent Manufacturing and Production management
- Taiwan Semiconductor Manufacturing Company (TSMC)
- Institute for Information Industry (III)
(in collaboration with top semiconductor testing/assembly service providers)
- Retail operations management
- Wellcome (#3 supermarket chain in Taiwan) and Industrial Technology Research Institute
- Digital-twin and process control for steel and textile industry
- China Steel Corporation (CSC), #17 steel company worldwide
- Institute for Information Industry(III), (in collaboration with top textile companies)
- Intelligent warehouse and material handling system management
- Foxconn (#1 EMS provider worldwide)
- Taiwan Semiconductor Manufacturing Company (TSMC)
Given recent development in sensor and communication technologies, abundant data can now be collected in all service and manufacturing systems; however, using those data effectively to enhance manufacturing and business decisions remains challenging. The combination of knowledge extraction and stochastic optimization will therefore be crucial for the success of any company in future business/manufacturing environments enabled by sensors and networks.