The situation in the semiconductor production lines change from time to time, requiring extemporal decisions. Under such circumstances, the decision-making method of production management has been developed based on the decision support system approach, which utilises an information system based on partial data generated by real time line and managers’ experience and intuition. We propose a system designed through deep learning method, to construct an optimal decision system for yield improvement. The aim of this study is to propose a system design that can support decision-making for yield improvement by using manufacturing deep neural network method in the semiconductor industry. The semiconductor manufacturing process data used is the production data consisting of 1,000 lots and 100 processes. The model developed in this system proposes decision variables with optimal yields within the constraint condition and supports the decision-making in the semiconductor production process by using the corresponding decision variables.
