臺灣博碩士論文加值系統

深度學習是實現人工智慧的技術，它開創了許多人工智慧的實際應用。其優勢在於它們比起動態或物理模型需要更少的開發時間，且相對地簡單。
本研究使用人工智慧類神經網路的一種計算技術，Echo State Network( ESN )演算法，是屬於Reservoir Computing( RC )的一種計算技術。它是一種簡單，快速和高效的非線性演算法，本研究使用該方法應用於雨量之預測以及測謊之判斷。
雨量是評估水資源、農業、生態系統及水文的重要依據。而使用深度學習演算法來預測是一項很有發展性的方式。本實驗使用 ESN 架構，並對台南地區曾文觀測站(120.497E,23.219N)及高雄潮位站(座標為120.283E, 22.617N)自 2002 年至 2014 年每小時的氣象資料進行分析。並將其與其他神經網路演算法做比較，結果表明 ESN的計算結果優於 MATLAB 類神經網路 toolbox 所提供的數種計算工具。此研究成果對於水庫發電與土石流防災有幫助。
本研究另一主題則是使用深度學習進行測謊的研究。目前普遍的測謊儀是藉由觀測受試者之皮膚電阻、呼吸波與脈搏波(血壓)等三項主要參數來測量人們的心理變化，而本研究則是將受試者之回答錄音，經過在語音辨識( Speech Recognition )以及語者辨識( Speaker Recognition )中常用到的特徵參數，稱為梅爾倒頻譜係數( Mel-scale Frequency Cepstral Coefficients, MFCC )之處理後再經過 ESN 架構進行演練，結果證明使用深度學習在測謊應用上也可以達到不錯的成效，平均正確率可達65%，最高正確率可高達100%，為測謊提供一個新的可行方式。

Based on artificial intelligence deep learning has been applied on many applications. The advantage lies in its less development time consumption and relatively more simple than the dynamic or physical models.
This study uses an easy, fast and efficient non-linear algorithm, Reservoir Computing ( RC ), based on Echo State Network ( ESN ) algorithm. We apply this algorithm for the prediction of rainfall and the deception detection.
Precipitation is a useful information for assessing vital water resources, agriculture, ecosystems and hydrology. Data-driven model predictions using deep learning algorithms are a promising way for these purposes. In this study, we used ESN to analyze the meteorological hourly data from 2002 to 2014 at the Tainan Zengwen Observatory
(120.497E,23.219N) and the sea level station in Kaohsiung (120.283E, 22.617N). We also compared the prediction and observation by using the ESN algorithm and a commercial neuron network MATLAB toolbox. The results show that the ESN can provide a better performance to predict rainfall.
Another subject of this study is the realization of deception detection based on deep learning. Nowadays, polygraph is performed by the observation of the subject's skin resistance, respiratory wave and pulse wave (blood pressure) to measure people's psychological changes. In this study, the vocal signal of the answer to the subjects is acquired. After the processing of the Mel-scale frequency coefficient (MFCC), which is commonly used in speech recognition and speaker recognition, the signal is treated by ESN. The result shows that the method applied on deception detection can also provide a good result. The average accuracy of the deception detection is 65%. The highest correctness can be up to 100%.

摘要 I
Abstract III
致謝 V
目錄 VI
圖目錄 IX
表目錄 XI
第一章、緒論 1
1.1 研究動機 1
1.2 人工智慧歷史回顧 1
1.3 人工智慧應用於雨量預測與測謊 4
1.4 結論 5
第二章、Reservoir Computing 6
2.1 Echo State Network 8
2.2 結論 11
第三章、深度學習演算法應用於台南曾文氣象站之雨量預測 12
3.1 研究動機及文獻回顧 12
3.2 資料蒐集 14
3.3 資料預處理 15
3.4 實驗流程 17
3.5 校驗方法 18
3.6 預測結果與討論 20
3.7 結論 28
第四章、深度學習演算法應用於測謊 29
4.1 研究動機及文獻回顧 29
4.2 資料蒐集 31
4.3 資料預處理 33
4.4 實驗流程 38
4.5 校驗方法 40
4.6 判斷結果與討論 40
4.7 結論 41
第五章、結論與未來工作 42
5.1 總結 42
5.2 未來工作 43
參考文獻 44

[1] https://www.bnext.com.tw/article/42632/what-is-ai
[2] https://www.bnext.com.tw/article/38923/BN-2016-03-14-120814- 178
[3] M. Lukoševičius and H. Jaeger, ”Reservoir computing approaches to
recurrent neural network training,” Computer Sci. Rev. 3, 127, 2009.
[4] L. Larger, A. Baylón-Fuentes and R. Martinenghi, ”High-Speed Photonic Reservoir Computing Using a Time-Delay-Based Architecture:Million Words per Second Classification,” Phys. Rev. X 7 , 011015, 2017.
[5] 張育誠, “Prediction of Sensory Input Using Echo State Network,” 碩士論文, 國立中正大學電機工程研究所, 2011.
[6] M. Lukoševičiusin, ”Neural networks: Tricks of the trade,” Künstliche
Intelligenz manuscript, 365, 2012.
[7] B. Schrauwen, D. Verstraeten and J. Campenhout, “An overview of
reservoir computing: theory, applications and implementations,”
Proceedings of the 15th European Symposium on Artificial Neural Networks, 471, 2007.
[8] C. Gallicchio and A. Micheli, “Deep reservoir computing: A critical
analysis,” European Symposium on Artificial Neural Networks,
Computational Intelligence and Machine Learning, 27, 2016.
[9] H. Jaeger, “Tutorial on training recurrent neural networks, covering
BPPT, RTRL, EKF and the echo state network,” GMD Report 159, 2002.
[10] H. Jaeger, “The “echo state” approach to analyzing and training
recurrent neural networks,” GMD Report 148, 2001.
[11] L. Richardson, ”Weather Prediction by Numerical Process,”
Cambridge, The University press, 1922.
[12] https://wol.jw.org/cmn-Hant/wol/d/r24/lp-ch/102001246
[13] 吳宗堯及葉信良, ”現有颱風預報研究成果作業化之研究(二) ,” 行政院國家科學委員會防災科技研究報告78-29號, 147, 1989.
[14] 申雍及陳守泓, ”梅雨期間作物承受豪雨風險機率之估算,” 中華農學會報 168, 93, 1994.
[15] 廖浩彥, ”利用雷達觀測直接反演氣象變數進行資料同化已改進短期定量降水預報-2008 SoWMEX IOP8 個案分析,” 碩士論文, 中央大學大氣物理研究所, 2014.
[16] 吳明進, 陸雲及童慶斌, ”區域氣候變遷模擬系統之整合與應用-子計畫VI：全球氣候變遷對台灣區域氣候與水資源衝擊之評析(I) ,” 行政院國家科學委員會, 2002.
[17] R. Deo and M. Şahin, “Application of the extreme learning machine algorithm for the prediction of monthly Effective Drought Index in eastern Australia,” Atmospheric Research Vol 153, 512, 2015.
[18] R. Hashim, ”Selection of meteorological parameters affecting rainfall estimation using neuro-fuzzy computing methodology,”
Atmospheric Research Vol 171, 21, 2016.
[19] P. Nayak, K. Sudheer and K. Ramasastri, “Fuzzy computing based rainfall–runoff model for real time flood forecasting,” Hydrological Processes Vol 19, 955, 2005.
[20] P. Nayak, K. Sudheer and D. Rangan, “A neuro-fuzzy computing technique for modeling hydrological time series,” Journal of Hydrology Vol 291, 52, 2004.
[21] 段智懷, ”倒傳遞類神經網路小區域颱風降雨預報-前饋式與遞迴式之比較,” 碩士論文,逢甲大學水利工程研究所, 2004.
[22] R. Teschl, W. Randeu and F. Teschl, ”Improving weather radar estimates of rainfall using feed-forward neural networks,” Neural networks Vol 20, 519, 2007.
[23] A. Manzato, A. Cicogna and A. Pucillo, ”6-hour maximum rain in Friuli Venezia Giulia:Climatology and ECMWF-based forecasts,” Atmospheric Research Vol 169, 465, 2016
[24] S. Sodoudi, A. Noorian and M. Geb, ”Daily precipitation forecast of ECMWF verified over Iran,” Theor Appl Climatol Vol 99, 39, 2010.
[25] H. Ruigar and S. Golian, ” Prediction of precipitation in Golestan dam watershed using climate signals,” Theor Appl Climatol Vol 123, 671, 2016.
[26] M. French, W. Krajewski and R. Cuykendall, ”Rainfall forecasting in space and time using a neural network,” Journal of Hydrology Vol 137, 1, 1992.
[27] W. Sun, S. Shan and C. Zhang, ”Prediction of Typhoon Losses in the South-East of China Based on B-P Network,” International Conference on Artificial Intelligence and Computational Intelligence, 2010.
[28] 吳明進, 陸雲及童慶斌, ”區域氣候變遷模擬系統之整合與應用-子計畫VI：全球氣候變遷對台灣區域氣候與水資源衝擊之評析(I),” 行政院國家科學委員會, 2002.
[29] 孫建平及張斐章, ”倒傳遞類神經網路演算法於時雨量預測之研究,” 農業工程研討會, 209, 1995.
[30] 張斐章, 胡湘帆及黃源義, ”反傳遞模糊類神經網路於流量推估之應用,” 中國農業工程學報, 第44卷, 第2期, 26, 1998.
[31] 林柏承, ”應用類神經網路於颱風降雨量的推估,” 碩士論文, 成功大學水利及海洋工程學系, 2000.
[32] 羅華強, ”類神經網路-MATLAB的應用,” 高立圖書, 2005.
[33] D. Smith, “Why We Lie,” St. Martin’s Press, 2004.
[34] B. DePaulo, D. Kashy and S. Kirkendol, ” Lying in Everyday Life,” Personality and Social Psychology Vol 70, 979, 1996.
[35] 徐國超, ”意在言外？--以口語化行為特徵進行『測謊鑑定前篩試驗』之可行性,” 碩士論文, 國立台北大學犯罪學研究所, 2011.
[36] 林故廷及翁景惠, ”測謊100問,” 書佑書局, 2003.
[37] National Research Council, ” The Polygraph and Lie Detection,“
The National Academic Press, 2003.
[38] C. Davatzikos, K. Ruparel and Y. Fan, ”Classifying spatial patterns of brain activity with machine learning methods: Application to lie detection,” NeuroImage Vol 28, 663, 2005.
[39] K. Park, H. Suk and H. Hwang, ”A functional analysis of deception detection of a mock crime using infrared thermal imaging and the Concealed Information Test,” Frontiers in Human Neuroscience, 7, 2013.
[40] Y. Zhou, H. Zhao and X. Pan, ” Deception detecting from speech signal using relevance vector machine and non-linear dynamics features,” Neurocomputing Vol 151, 1042, 2015.
[41] 王小川, “語音訊號處理,” 全華圖書股份有限公司, 2009.
[42] 陳奕宏, ”32位元處理器之定點數MFCC演算法的改進與探討,” 碩士論文, 國立清華大學資訊工程學系, 2006.
[43] J. Roger, "Audio Signal Processing and Recognition," (in Chinese) available at the links for on-line courses at the author's homepage at
http://www.cs.nthu.edu.tw/~jang.
[44] 謝依蘭, ”語音訊號數位處理,” 松崗電腦圖書資料股份有限公司, 1992