人工智慧是未來科技發展的趨勢，其中的技術就是機器學習，又深度學習是機器學習最重要的分支，約三十年前就有專家提出類神經網路的概念，此後不斷地有專家投入研究，但由於運算、學習速度的限制，一直不受到重視，直到 2010 年前後GPU效能逐漸強化且普及、使得平行運算速度加快。加上現代大數據的影像、文字、數據如洪水般湧出、工業4.0 需要大量數據運算、機器人取代人力及自駕車等，在這樣各種環境及需求搭配的情況下，出現爆發性的成長。但殺手應用尚未出現，所以技術仍在萌芽的階段仍值得投入研究、找出未來趨勢及方向。
本研究有別於傳統的方法分析新興產業，從專利管理圖中得知主要領導公司，並找出其領導公司主要發展的技術分類及技術，進行公司競爭力分析。並進行專利指標計算、探討公司各自主要發展之技術的專利品質。本研究結果可以做為新創可能投入的技術研發或創新方向。
研究結果得知，深度學習在專利部份仍為演算法、圖形辨識、語音辨識、數據計算、醫療及量測等應用最為廣泛，但其中又以演算法、圖形辨識及語音辨識相關專利間的競爭最為激烈，可見相關應用如機器視覺、語言及語音辨識仍是各家主力發展的方向。

Artificial intelligence is the future of scientific and technological development which the technology is machine learning. Deep Learning is the most important part of machine learning. Thirty years ago, some experts put forward the concept of neural network. Experts continue to put into research but limited by computing speed. Until 2010, GPU performance gradually enhanced and universal, making parallel computing speed. Recently big data, images, text growth fast, industrial 4.0 need robot replace the human and self-driving. In this kind of environment and demand with the circumstances, the explosive growth. But the killer application has not yet appeared, so the technology is still in the bud stage is still worthy of study to identify future trends and direction.
This research is different from the traditional methods of analysis of new industries. This research is study from the patent management map to get the main leading companies and to identify the leading companies to develop the main technical classification and competition. Calculating patent parameters of leading company's main development to find the technology of the patent quality. The results of this research can be used as a new R & D or innovation and provide invest direction.
The results show, deep Learning is still the most widely used in the patent. Such as algorithm, pattern recognition, speech recognition, data calculation, medical treatment and measurement. The most intense competition were pattern recognition and speech recognition. Visible applications were machine vision, natural language and voice recognition still the main direction of development.

(一) Books:
1. Ethem Alpaydin (2004).Introduction to Machine Learning.
2. IBM Corp. (2012). IBM SPSS Statistics for Windows，Version 21.0.
(二) 論文:
1. A Krizhevsky, I Sutskever, GE Hinton, (2012). Imagenet classification with deep convolutional neural networks.
2. Y Jia, E Shelhamer, J Donahue, S Karayev…, (2014). Caffe: Convolutional architecture for fast feature embedding.
3. Y LeCun, B Boser, JS Denker, D Henderson…, (1989). Backpropagation applied to handwritten zip code recognition.
4. R Girshick, J Donahue, T Darrell…, (2014). Rich feature hierarchies for accurate object detection and semantic segment.
5. K Simonyan, A Zisserman, (2014). Very deep convolutional networks for large-scale image recognition.
6. GE Hinton, N Srivastava, A Krizhevsky…, (2012).Improving neural networks by preventing co-adaptation of feature detectors.
7. C Szegedy, W Liu, Y Jia, P Sermanet…, (2015). Going deeper with convolutions.
8. BB Le Cun, JS Denker, D Henderson…, (1990). Handwritten digit recognition with a back-propagation network.
9. MD Zeiler, R Fergus, (2014).Visualizing and understanding convolutional networks.
10. N Srivastava, GE Hinton, A Krizhevsky…, (2014). Dropout: a simple way to prevent neural networks from overfitting.
11. P Sermanet, D Eigen, X Zhang, M Mathieu…, (2013). Overfeat: Integrated recognition, localization and detection using convolutional networks.
12. A Krizhevsky, G Hinton, (2009). Learning multiple layers of features from tiny images.
13. J Donahue, Y Jia, O Vinyals, J Hoffman, N Zhang…, (2014). DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition.
14. Y Taigman, M Yang, MA Ranzato…, (2014). Deepface: Closing the gap to human-level performance in face verification.
15. R Salakhutdinov, GE Hinton, (2009). Deep Boltzmann Machines.
16. Y LeCun, Y Bengio, (1995). Convolutional networks for images, speech, and time series.
17. A Sharif Razavian, H Azizpour, J Sullivan…, (2014). CNN features off-the-shelf: an astounding baseline for recognition.
18. C Farabet, C Couprie, L Najman…, (2013). Learning hierarchical features for scene labeling.
19. J Long, E Shelhamer, T Darrell, (2015). Fully convolutional networks for semantic segmentation.
20. IJ Goodfellow, D Warde-Farley, M Mirza, AC Courville…, (2013). Maxout networks.
21. K Chatfield, K Simonyan, A Vedaldi…, (2014). Return of the devil in the details: Delving deep into convolutional nets.
22. A Karpathy, G Toderici, S Shetty, T Leung…, (2014). Large-scale video classification with convolutional neural networks.
23. A Karpathy, L Fei-Fei, (2015). Deep visual-semantic alignments for generating image descriptions.
24. K He, X Zhang, S Ren, J Sun, (2015). Delving deep into rectifiers: Surpassing human-level performance on imagenet classification.
25. M Oquab, L Bottou, I Laptev, J Sivic, (2014). Learning and transferring mid-level image representations using convolutional neural networks.
26. Y LeCun, K Kavukcuoglu, C Farabet, (2010). Convolutional networks and applications in vision.
27. B Zhou, A Lapedriza, J Xiao, A Torralba…,(2014). Learning deep features for scene recognition using places database.
28. K He, X Zhang, S Ren, J Sun, (2014). Spatial pyramid pooling in deep convolutional networks for visual recognition.
29. J Donahue, L Anne Hendricks…, (2015). Long-term recurrent convolutional networks for visual recognition and description.
30. K Simonyan, A Zisserman, (2014). Two-stream convolutional networks for action recognition in videos.
31. Sun Y, Wang X, Tang X. (2014). Deep learning face representation from predicting 10,000 classes.
32. Rafal Jozefowicz, (2014). An Empirical Exploration of Recurrent Network Architectures.
33. Peter Foldiak and Dominik Endres, (2008). Sparse coding.
34. 袁建中, (2014). 美國非專利實施實體的涉訟專利特徵之研究, 博士論文，雲林科技大學
(三) 期刊:
1. (Chang et al, 2012)
Chang, K.-C., Chen, D.-Z. and Huang, M.-H. (2012). The relationships between the patent performance and corporation performance. Journal of Informetrics 6,no. 1: 131-39.
2. (Zhang et al. 2012)
Zhang, S., Yu, C.C., Chang, K.C. and Ken, Y. (2012). Exploring the nonlinear effects of patent h index, patent citations, and essential technological strength on corporate performance by using artificial neural network. Journal of Informetrics 6, no. 4: 485-95
3. 楊牧民, 陳達仁, & 黃慕萱，「從美國專利看台灣企業科技創新競爭力」，政大智慧財產評論，200410 (2:2期)，2004，1-24。
4. Autoencoder for Words, Liou, C.-Y., Cheng, C.-W., Liou, J.-W., and Liou, D.-R., Neurocomputing, Volume 139, 84–96, 2014
5. (Tong, and Frame, 1994)
Tong, X. and Frame, J. D. (1994). Measuring national technological performance with patent claims data. Research Policy, vol. 23, pp. 133–41.
6. Schankerman, M. (1998). How valuable is patent protection? Estimates by technology field. RAND Journal of Economics 29 (1), 77–107.
7. Schankerman, M. and Pakes, A. (1986). Estimates of the Value of Patent Rights in European Countries During the Post-1950 Period. The Economic Journal, 96(384): 1052-76.
8. Lanjouw, J.O., Ariel, P. and Jonathan, P. (1998). How to Count Patents and Value Intellectual Property: The Uses of Patent Renewal and Application Data. The Journal of Industrial Economics, 46(4): 405-32.
9. Lanjouw, J.O. and Mark, S. (2001). Characteristics of Patent Litigation: A Window on Competition. RAND Journal of Economics (2001), 129-151.
10. Lanjouw, J.O. and Mark, S. (2001). Patent Quality and Research Productivity: Measuring Innovation with Multiple Indicators. The Economic Journal, 114(495): 441-65.
(四) 研究報告:
1. (Tractica, 2016)
How Deep Learning Is Enabling Computer Vision Markets, Tractica, Bruce Daley, May 2, 2016
2. (Goldman Sachs Global Investment Research, 2016)
Artificial Intelligence, Goldman Sachs Global Investment Research, 2016.
(五) Website:
1. (周秉誼, 2016)
http://www.cc.ntu.edu.tw/chinese/epaper/0038/20160920_3805.html
2. (Michael Copeland, 2016)
https://blogs.nvidia.com.tw/2016/07/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/
3. (Yann LeCun, 2014)
http://cilvr.cs.nyu.edu/doku.php?id=deeplearning:slides:start
4. (Peter Morgan, 2016)
https://www.slideshare.net/SheamusMcGovern/deep-learning-frameworks-slides
5. (Geoffrey Hinton, 2016)
https://zh-tw.coursera.org/learn/neural-networks
6. http://ufldl.stanford.edu/tutorial/unsupervised/Autoencoders/
7. (Edwin Chen, 2011)
http://blog.echen.me/2011/07/18/introduction-to-restricted-boltzmann-machines/
8. (Hinton, 2011)
https://agollp.wordpress.com/2013/12/29/deep-belief-network/
9. (Arno Candel, 2016)
http://www.liuhaihua.cn/archives/263216.html
10. (Xuedong Huang, 2016)
http://www.technews.cn/2016/01/28/microsoft-releases-cntk-its-open-source-deep-learning-toolkit/
11. (Siemens.com, 2017)
https://www.siemens.com/global/en/home/company/about.html
12. https://kknews.cc/zh-tw/tech/5pr25l.html
13. https://Sony.net
14. (王明德, 2016)
https://www.ctimes.com.tw/DispArt/tw/物聯網/人工智慧/富士通/感測器/機器人/16062011077P.shtml
15. (Cho Mu-Hyun, 2016)
http://www.zdnet.com/article/samsung-applies-deep-learning-to-ultrasound-device/
16. http://www.nec-labs.com/research-departments/machine-learning/machine-learning-projects
17. https://zh.wikipedia.org/wiki/西安?子科技大?
18. (Pallav Sarma, 2016)
https://www.spe.org/en/twa/twa-article-detail/?art=2561
19. (Tim Cheng, 2017)
http://www.gereports.com/ge-reports-4q-earnings-highlights/
20. http://pl.mitsubishielectric.com/en/news-events/releases/2016/1014-a/index.page
21. https://www.qualcomm.com/news/releases/2016/05/02/qualcomm-helps-make-your-mobile-devices-smarter-new-snapdragon-machine
22. (Yoshua Bengio, 2016)
http://www.digitimes.com.tw/tw/dt/n/shwnws.asp?id=0000464105_OXL2WTETLHI57K6L25WGC&pack=10308&cnlid=1&cat=10