(3.236.222.124) 您好!臺灣時間:2021/05/14 16:32
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果

詳目顯示:::

我願授權國圖
: 
twitterline
研究生:Dumsile Mhlanga
研究生(外文):Dumsile Mhlanga
論文名稱:A Study of Machine Learning and its Application on Miniature Spectrometer
論文名稱(外文):機器學習之研究和應用於微型光譜儀
指導教授:Cheng-Chun Chang
口試委員:Cheng-Chun Chang, Jui-Ching Cheng, Yu-Fu Hsieh
口試日期:2016-07-21
學位類別:碩士
校院名稱:國立臺北科技大學
系所名稱:電資學院外國學生專班
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2016
畢業學年度:104
中文關鍵詞:Miniature SpectrometerMachine LearningArtificial Neural Network AlgorithmFeedforward Neural NetworkBack-Propagation Neural Network
外文關鍵詞:Miniature SpectrometerMachine LearningArtificial Neural Network AlgorithmFeedforward Neural NetworkBack-Propagation Neural Network
相關次數:
  • 被引用被引用:0
  • 點閱點閱:95
  • 評分評分:
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
Recently there has been renewed interest on machine learning. This is due to its powerful modelling ability as well as the existence of some efficient learning algorithms. A prominent example of such algorithms is multilayer learning machine, which is designed to train features of higher levels by applying the composition of lower level features. The information to be learned is not located in a single layer of neurons but distributed representation allows deep learning networks to have a strong capacity for learning and produce much better generalization when compared to the traditional machine learning.
In this paper we introduce machine learning algorithms with its application on miniature spectrometers, which due to its non-ideal property reconstruction methods are required. Experiments from machine learning algorithms shows that a significantly of accurate model that can be used for future prediction with its ability to autonomously generate high-level representations from raw data source can be achieved. When multi-layer neural network is used to train a miniature spectroscopic raw data, the output spectrum is taking a form of a target spectrum. Finally, when comparing the single and multi-layer neural networks algorithms on the raw spectrum produce by the miniature spectrometer to develop a model and the multi-layer neural network show it achieves on the reduced error estimation over the single layer neural network.
Recently there has been renewed interest on machine learning. This is due to its powerful modelling ability as well as the existence of some efficient learning algorithms. A prominent example of such algorithms is multilayer learning machine, which is designed to train features of higher levels by applying the composition of lower level features. The information to be learned is not located in a single layer of neurons but distributed representation allows deep learning networks to have a strong capacity for learning and produce much better generalization when compared to the traditional machine learning.
In this paper we introduce machine learning algorithms with its application on miniature spectrometers, which due to its non-ideal property reconstruction methods are required. Experiments from machine learning algorithms shows that a significantly of accurate model that can be used for future prediction with its ability to autonomously generate high-level representations from raw data source can be achieved. When multi-layer neural network is used to train a miniature spectroscopic raw data, the output spectrum is taking a form of a target spectrum. Finally, when comparing the single and multi-layer neural networks algorithms on the raw spectrum produce by the miniature spectrometer to develop a model and the multi-layer neural network show it achieves on the reduced error estimation over the single layer neural network.
Table of Contents
Abstract..........................................ii
Acknowledgement..................................iii
Table of Contents.................................iv
List of Tables....................................vi
List of Figures..................................vii
Chapter 1 Introduction.............................1
1.1 Motivation.................................1
1.2 Overview...................................2
1.3 Structure of the Thesis....................2
Chapter 2 Background and Related Work..............3
2.1 Machine Learning..........................3
2.1.1 Background of Machine Learning......3
2.1.2 Artificial Neural Network...........6
2.2 Vis/NIR Spectroscopy......................9
2.3 Miniature Spectrometers..................15
Chapter 3 Machine Learning and Problem Formulation.......................................18
3.1 Single Layer Neural Network..............18
3.1.1 Feed-forward Neural Network........18
3.1.2 Back Propagation Neural Network....20
3.2 Multi-Layer Neural Network...............22
3.3 Future Perspective of Neural Network.....26
3.3.1 Deep Learning Algorithms...........26
3.3.2 Three Classes of Deep Learning
Network............................28
3.3.3 Deep Networks for Supervised
Learning...........................29
3.3.3.1 General Issues of Supervised
Learning Algorithms........30
Chapter 4 Simulations Results and Performance Analysis..........................................32
4.1 Data Collection..........................32
4.2 Simulation Setup for Neural Network......34
4.3 Deriving a Model Using ANN/MNN...........35
4.4 Analysis of Simulation Results on Single
Layer Neural Network.....................38
4.5 Analysis of Simulation Results on Multi-layer Neural Network..............................42
4.6 Comparison between Single-Layer and Multi-
Layer Neural Network ....................48
Chapter 5 Conclusion and Future Works…...........50
References........................................52
Terminology.......................................55
[1] F.-C. Tsui, "Technical description of rods: a real-time public health," american medical informatics association, vol. 10, pp. 399-408, 2003.
[2] E. Ackerman, "IEEE spectrum," 20 May 2016. [Online]. Available: http://spectrum.ieee.org/automaton/robotics/artificial-intelligence/watchbot-takes-care-of-absentminded-humans. [Accessed 7 July 2016].
[3] S. T. G. D. R. &. M. S. G. John Darrell Van Horn, "Sharing neuroimaging studies of human cognition," natural neuroscience, vol. 7, pp. 473-481, 2004.
[4] V. A. G. Hinton, "Rectified linear units improve restricted botztmann machine," in Proceedings of the 27th International Confrence Machine Learning, 2010.
[5] E. M. Agency, "Guideline on the use of near infrared spectroscopy (NIRS) by the pharmaceutical industry and the data requirements for new submission and variations," 2012. [Online].
[6] C.Pasquini, "Near infrared spectroscopy: Fundamentals, practical aspect and analytical applications," J.Braz.Chem, vol. 14, April 2002.
[7] P. C., "Near infrared spectroscopy: Fundamentals, Practical Aspects and Analytical Applications," 2003.
[8] I. V. M. Blanco, "NIR spectroscopy: A rapid-respond analytical tool," 2002.
[9] F. J. H. A. T. A. N. D. A. Skoog, "Principios_Del_Anlisis_Instrumental_Apendices".
[10] E. A. H. C. C.N. Banwell, Fundamentals of molecutar spectroscopy, F. Edit, Ed., McGraw Hill Education, 2013.
[11] I. Barton, "FE theory and principles of near infrared spectroscopy," Spectrosc.EUR, vol. 14, no. 1, pp. 12-18, 2002.
[12] Hamamatsu, february 2016. [Online]. Available: http://files.instrument.com.cn/17img/files/201511/attachment/45d72fe4-89b2-439c-84f9-774cbde892ea.pdf. [Accessed 24 july 2016].
[13] Mazet, "Mazet datasheet," Electro optics, Netherlands, 2016.
[14] Y. Q. W. a. X. Q. Bai, "Prediction and evaluation of water quality using BP neural network," Journal of Wuhan Polytechnic University, vol. 26, no. 1, pp. 64-67, 2007.
[15] R. .. Erb, "Introduction to backpropagation neural network computation," Pharm. Res, no. 10, p. 165–170., 1993.
[16] R. Hecht-Nielsen, "Kolmogorov’s mapping neural network existence theorem," in Proc. First IEEE Int. Joint Conf.Neural Networks, San Diego, 1987.
[17] D. F. M .N. Jadid, "Predicting moment-curvature parameters from experimental data," Eng. Appl. Artif. Intell, no. 9, p. 303–319, 1996.
[18] R. D. J . Sietsma, "neural net pruning - why and how,," in IEEE Int. Conf. Neural Networks, 1988.
[19] L. D. A. D. Yu, "Deep learning methods and applications," Foundations and Trends in Signal Processing, vol. 7, no. 3-4, 2013.
[20] Y. B. Yann LeCun, "Review deep learning," Nature, vol. 521, pp. 436-444, 28 May 2015.
[21] S. Z. C. Y. Q. Zhang, "Data preparation for data mining," Applied Artificia lntelligence, vol. 17, pp. 375 - 381., 2002.
[22] G. &. M. M. Batista, "An analysis of four missing data treatment methods for supervised learning," Applied Artificial Intelligence, vol. 17, pp. 519-533, 2003.
[23] H. A. H. M. Liu, "Instance selection and constructive data mining," Kluwer, Boston, 2001.
[24] R. T., "A unifying view on instance selection," Data Mining and Knowledge Discovery, vol. 6, p. 191–210, 2002.
[25] B. Wythoff, "Backpropagation neural networks: a tutorial,," Intell. Lab. Syst 18, p. 15–155., 1993.
[26] J. G. J . Zupan, "Neural networks for chemists:an introduction," VCH, New York, 1993.
[27] M. H. I .A. Basheer, "Artificial neural networks: fundamentals, computing, design, and application,," J. Microbiol. Methods, no. 43, p. 3–31., 2000.
[28] X. A. Y. Bengio, "Understand the difficults of traning deepfeedfoward neural network," Proceeding of Artificial Intelligence and Statistics (AISTATS), 2010.
[29] B. I. Choi, C.-C. Chang and U. Kurokawa, "Filter-based miniature spectrometers: Spectrum reconstruction using adaptive regularization," IEEE SENSORS JOURNAL, pp. VOL. 11, NO. 7, JULY 2011.
[30] G. H. A. R. Salakhutdinov, "Reducing the dimensionality of data with neural network," in Science.
[31] Y. B. and O. Delalleau, "On the expressive power of deep architectures," in Algorithmic Learning Theory, 2011,.
[32] Y. Bengio, "Learning deep architectures for AI," in Foundations and Trends in Machine Learning, 2009.
[33] I. S. A. G. E. H. A. Krizhevsky, "mageNet classification with deep convolutional neural networks," in Advances in Neural Information Processing Systems, 2012.
[34] G. E. H. A. R. J. W. D. E. Rumelhart, "Learning representations by back-propagating errors,," in Nature, 1986.
[35] M. N. H. A. I. A. Basheer, "A hybrid bayesian-neural network approach for probabilistic modeling of bacterial growth/no-growth interface," in International Journal of Food Microbiology, 2003..
[36] J. Coates, "A review of new small-scale technologies for near infrared measurements," 18 June 2014.
[37] E. M. Agency, "Guigeline on the use of near infrares spectroscopy (nirs) by the pharmaceutical industry and the date requirements for new submission and variations," 2012. [Online]. Available: http://www.EMEA/CVMP/QWP/17760/2006REV2.
[38] W. F. McClure, "Near-infrared spectroscopy—The giant is running strong," Anal.Chem., pp. A43-A53, 1994.
[39] M. B. A. I. Villarroya, "NIR Spectroscopy: A Rapid-response analytical tool," vol. 21, no. 4, pp. 240-250, April 2002.
[40] C.-C. Chang, Heung-No Lee, "On the estimation of target spectrum for filter-array based spectrometers," in Electrical and Computer Engineering, University of Pittsburgh,, Pittsburgh PA,USA, 20o8.
連結至畢業學校之論文網頁點我開啟連結
註: 此連結為研究生畢業學校所提供,不一定有電子全文可供下載,若連結有誤,請點選上方之〝勘誤回報〞功能,我們會盡快修正,謝謝!
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top