網際網路以及資料存儲科技快速發展，使得企業可以更有效的收集數據，然而，龐大的資料如果沒有進行適當的分析就沒有辦法為企業帶來價值。近年來，資料科學的發展大大的幫助人類從巨量的資料中找出關鍵的資訊，類神經網路有著建構複雜模型的能力，在半導體產業進步的推波助瀾之下，類神經網路已經可以有效率的完成預測，並且可以應用於製造、醫療、行銷等專業領域。
對製造業而言，需要滿足顧客的往往不只要有良好的生產品質，更需要有良好的生產彈性以應付顧客多變的需求。本篇研究討論如何使用類神經網路以及基因演算法幫助決定半導體生產的製程參數，藉由類神經網路建立的模型作為基因演算法目標函式，就能夠發揮基因演算法強大的搜尋能力。在本篇的個案研究中，此方法幫助工程師快速地找出製程參數的組合，並且成功的製造出100%良率以及晶圓內徑的結果極為接近目標值的產品。相較於使用業界常用的實驗設計、田口方法以及反應曲面方法，類神經網路結合基因演算法的方式只需要利用過去的歷史資料即可找出最佳的參數組合，在有著更高的準確率的情況下，此方法更能夠有效的降低實驗進行的次數，進一步的幫助企業降低成本以及增加生產彈性。

The rapid development of the Internet and data storage technologies has allowed companies to collect data more effectively. However, data cannot bring value to companies without proper analysis. In recent years, the development of data science has greatly helped humans to find key information from big data. Neural networks have the ability to construct extremely complex models. With the help of the progress of the semiconductor industry, neural networks have the ability to forecast very efficiently, and it can be applied to manufacturing, medical, marketing, and other professional fields.
For the manufacturing industry, it is not only necessary to have good production quality, but also to have good production flexibility to meet the changing needs of customers. This research discusses how to use neural networks and genetic algorithm to help determine the process parameters of semiconductor production. With the use of neural networks as the target function, genetic algorithm is able to search the optimal solution with collected data. In the case study, our proposed method can help engineers quickly find the best combination of process parameters, and successfully manufacture products with 100% yield and wafer inner diameter that is very close to the target value. Compared with the commonly used experimental design, Taguchi method, and response surface method, our proposed method only needs to collect historical data. However, the results showed that it has higher accuracy. In the case of higher accuracy, this method can effectively reduce the number of experiments conducted, further help companies reduce costs and increase production flexibility.

1. Bengio Y., Goodfellow I. and Courville A. (2016), Deep Learning
2. Brynjolfsson E. and McAfee A. (2012), “Big Data: The Management Revolution,” Harvard business review, 90(10), pp. 60-68
3. Fausett L. V. (1994), Fundamental of Neural Network: Architecture, Algorithms, and Application
4. Gaidhane R., Vaidya C. and Raghuwanshi M. (2014), “Intrusion detection and attack classification using back-propagation neural network,” International journal of engineering research and technology, Vol. 3
5. Glorot X. and Bengio Y. (2010), “Understanding the difficulty of training deep feedforward neural networks,” Journal of Machine Learning Research, pp. 249-256.
6. Goldberg D. E., (1989), Genetic algorithms in search, optimization, and Machine Learning
7. Gröger C., Florian N. and Bernhard M. (2012), “Data Mining-driven Manufacturing Process Optimization,” Conference: World Congress on Engineering (WCE)
8. George D., Suich R. (1980), “Simultaneous optimization of several response variables,” Journal of Quality Technology, Vol. 12, Issue 4
9. Jain A. K., Mao J. C. and Mohiuddin K. M. (1996), “Artificial Neural Networks: A Tutorial,” Computer 29.3, pp. 31-44
10. Kobayashi T. and Simon D. L. (2005), “Hybrid Neural-Network Genetic-Algorithm Technique for Aircraft Engine Performance Diagnostics,” Journal of propulsion and power, Vol. 21, No. 4,
11. Kotsiantis S. B., Kanellopoulos D. and Pintelas P. E. (2006), “Data Preprocessing for Supervised Learning,” International journal of computer science, Vol. 1, No. 1, pp. 111-117
12. Li D., Hinton G. and Kingsbury B. (2013), “New types of deep neural network learning for speech recognition and related applications: an overview,” IEEE Acoustics, Speech, and Signal Processing International Conference, pp. 8599-8603
13. Lin H. C., Su C. T., Wang C. C., Chang B. H. and Juang R. C. (2012), “Parameter optimization of continuous sputtering process based on Taguchi methods, neural networks, desirability function,” Expert Systems with Applications, Vol. 39, No. 17, pp. 12918-12925.
14. Nair V. and Hinton G. E. (2010), “Rectified Linear Units Improve Restricted Boltzmann Machines,” International Conference on Machine Learning, pp. 807–814
15. Panchal G., Ganatra A., Kosta Y. P. and Panchal D. (2011), “Behaviour analysis of multilayer perceptrons with multiple hidden neurons and hidden layers,” International Journal of Computer Theory and Engineering, Vol. 3, No. 2, pp. 332-337
16. Schmidhuber J. (2015), Deep learning in neural networks: An overview
17. Sette S., Boullart L., Langenhove L. V. and Kiekens P. (1997), “Optimizing the fiber-to-yarn production process with a combined neural network/genetic algorithm approach,” Textile Research Journal, Vol. 67, pp. 84-92
18. Su C. T., C. Lin M. and Chang C. A. (2012), “Optimization of the Bistability Property for Flexible Display by an Integrated Approach Using Taguchi methods, Neural Networks and Genetic Algorithms,” Microelectronics Reliability, Vol. 52, No. 17, pp. 12918-12925
19. Su C. T., Hsiao Y. H. and Chang C. C. (2012), “Parameter Optimization Design for Touch Panel Laser Cutting Process,” IEEE Transactions on Automation Science and Engineering, Vol. 9, No. 2, pp. 320-329
20. Tsai J. T., Chou J. H. and Liu T. K. (2006), “Tuning the structure and parameters of a neural network by using hybrid Taguchi-genetic algorithm,” IEEE Transactions on Neural Networks, Vol. 17, No. 1, pp. 69-80
21. Whitley D. (1994), “A genetic algorithm tutorial,” Statistics and Computing, Vol. 4, pp. 65–85
22. Widrow B., Winter R. G. and Baxter R. A. (1987), “Learn phenomena in layered neural networks,” Proc. 1st IEEE Intl. Conf. on Neural Networks, Vol 2, pp. 411-429