近幾年來，華語流行音樂的發展越來越多元，而大眾所接收到的資訊是流行音樂當中的組成元素”曲與詞”，兩者分別具有賦予人類感知的功能，使人能夠深刻體會音樂作品當中所表答的內容與意境。然而，作曲與作詞都是屬於專業的創作藝術，作詞者通常在填詞時，會先對樂曲當中的結構進行粗略的分析，找出整首曲子的曲式，而針對可以填詞的部份，再進行更細部的分析將詞填入最適當的位置。流行音樂當中，曲與詞存在著密不可分的關係，瞭解歌曲結構不僅能降低填詞的門檻，亦能夠明白曲子的骨架與脈絡;在音樂教育與音樂檢索方面亦有幫助。 本研究的目標為，使用者輸入流行音樂歌曲，系統會自動分析出曲子的『曲式結構』。方法主要分成三個部分，分別為主旋律擷取、歌句分段與音樂曲式結構擷取。首先，我們利用Support Vector Machine以學習之方式建立模型後，擷取出符號音樂中之主旋律。第二步驟我們以”歌句”為單位，對主旋律進行分段，對於分段之結果建構出Self-Similarity Matrix矩陣。最後再利用Non-Negative Matrix Factorization針對不同特徵值矩陣進行分解並建立第二層之Self-Similarity Matrix矩陣，以歧異度之方式找出曲式邊界。 我們針對分段方式對歌曲結構之影響進行分析與觀察。實驗數據顯示，事先將歌曲以歌句單位分段之效果較未分段佳，而歌句分段之評測結果F-Score為0.82;將音樂中以不同特徵值建構之自相似度矩進行Non-Negative Matrix Factorization後，另一空間中之基底特徵更能有效地分辨出不同的歌曲結構，其F-Score為0.71。 / Music structure analysis is helpful for music information retrieval, music education and alignment between lyrics and music. This thesis investigates the techniques of music structure analysis for Chinese popular music. Our work is to analyze music form automatically by three steps, main melody finding, sentence discovery, and music form discovery. First, we extract main melody based on learning from user-labeled sample using support vector machine. Then, the boundary of music sentence is detected by two-way classification using support vector machine. To discover the music form, the sentence-based Self-Similarity Matrix is constructed for each music feature. Non-negative Matrix Factorization is employed to extract the new features and to construct the second level Self-Similarity Matrix. The checkerboard kernel correlation is utilized to find music form boundaries on the second level Self-Similarity Matrix. Experiments on eighty Chinese popular music are performed for performance evaluation of the proposed approaches. For the main melody finding, our proposed learning-based approach is better than existing methods. The proposed approaches achieve 82% F-score for sentence discovery while 71% F-score for music form discovery.

本文建立小型開放經濟體系的VAR模型，利用符號限制法（Sign Restrictions）認定財政政策衝擊，評估台灣財政政策的總體經濟效果。符號限制法係利用對衝擊反應函數做符號限制的方式認定財政衝擊，對關心的變數如：實質GDP、民間消費、民間投資、貿易收支等變數對財政政策衝擊的反應則不設限制，讓資料來回答。本研究參酌Mountford and Uhlig（2009）及Ho and Yeh（2010）的方式認定總合供給衝擊、總合需求衝擊、貨幣政策衝擊、政府支出衝擊以及政府收入衝擊。研究結果發現，政府支出衝擊對民間投資短期會產生排擠效果，中長期（二十季）則有提振的效果。政府支出衝擊引發短期名目利率上漲，國外資金流入，實質有效匯率上升，貿易收支因而下跌。政府支出衝擊對於實質GDP一開始有正向效果，但排擠效果短期會使實質GDP下跌，一旦政府支出帶動中長期民間投資後，對實質GDP有正向效果，但並不顯著。 政府收入衝擊短期對實質GDP、民間消費、民間投資有正向效果，中長期的效果為負。若以政府支出衝擊細項來看，政府消費支出衝擊對實質GDP有顯著提振的效果，政府投資支出衝擊對於實質GDP的助益十分有限。

本研究係探討匯率與總體經濟之關聯性，以中國大陸1996第一季至 2013年第一季之總體經濟變數，共計樣本數為69筆季資料。先以1996第一季至 2013年第一季全期數據進行實證分析。再以2005年7月為分界點，分為1996年第一季至2005年第二季及2005年第三季至2013年第一季數據分別進行實證分析。 本論文就REER、GDP、CPI、M2、UNEMP、CHIBOR、FDI、OPEN等總體經濟變數，以單根檢定及建構向量自我迴歸模型進行實證分析，並以Granger因果關係檢定、衝擊反應分析及預測誤差變異數分解，以了解匯率與總體經濟相互間之關係。 實證結果發現，中國大陸匯率與總體經濟間的關係自2005年7月21日匯率改革後逐漸增強，但整體言之匯率與總體經濟間之傳導能力仍然不大，人民幣匯率的變動主要受其自身影響較多，受總體經濟變數的相互影響較小，顯示其外匯市場的開放程度與一個真正開放的經濟體還是有些許差距。 / This research examines the correlation between foreign exchange rates and macroeconomics by using the data of economic variables of China from the 1st quarter of 1996 to the 1st quarter of 2013. The sample contains 69 quarterly data during the entire period, while the reform of Chinese exchange rate on 21st July 2005 is a crucial division. In order to find the correlation between foreign exchange rates and macroeconomics, the research examines the economic variables such as REER, GDP, CPI, M2, UNEMP, CHIBOR, FDI, and OPEN by using unit root test, vector autoregression model, Granger causality test, impulse response function and variance decomposition impulse response function. The result of the tests indicates that after the reform of Chinese exchange rate on 21st July 2005, the correlation between exchange rates and macroeconomics has been enhanced, but the connection is not prominent. In other words, the fluctuation of Renminbi is mainly affected by the nation’s policy instead of its macroeconomic factors. Hence, the openness of the Chinese foreign exchange market is still distant from a real open economy.

本篇探討在2010年4月16日滬深300股指期貨正式上市到2012年9月18日止的連續近月每日收盤日資料，進行區域內金融期貨市場連動關係的研究，試圖發現兩岸三地之股價指數期貨市場在亞太地區的金融主導地位，以作為國內外投資者在區域內的投資決策參考。 實證結果顯示，從共整合及向量誤差修正模型檢定發現，兩岸三地股指期貨具有長期均衡及短期的互動關係，因此可以視此三地為單一區域市場。在Granger因果檢定上，台股指數期貨雖無法預測恆生指數期貨，但仍明顯領先滬深300股指期貨且程度大於恆生指數期貨，或可推測兩岸因ECFA的簽訂使實體經濟的關聯性更為緊密，至於恆生指數期貨大多以金融、地產股為其主要成分，與大陸主要以實體經濟為主的金融市場，其Granger預測滬深300股指期貨的能力因此相對較弱。另由衝擊反應檢定得知恆生指數期貨為一獨立的市場，不受台灣及大陸指數期貨市場衝擊的影響；滬深300指數期貨因大陸金融市場逐漸開放，也會受到香港及台灣金融期貨市場之衝擊而產生影響；至於台股指數期貨則在兩岸三地，最易受到其他市場影響。最後由預測變異數分解檢定發現，台股指數期貨及滬深300股指期貨的波動皆易受到恆生股價指數期貨變異的影響，而恆生指數期貨在兩岸三地間之解釋能力最強，於兩岸三地間具金融主導地位。至於台股指數期貨對大陸金融期貨的影響也有突出的表現，因此若政府有心推展亞太金融中心之營運，勢必得加強區域間整合的力度，提出有利吸引外資之最政策，以增加台灣股市於國際間之競爭力。 / This study conducts analysis of regional linkage between financial futures market by examining consecutive daily closing information from April 16, 2010 (the official list date of CSI 300 index futures) to September 18, 2012. This study tries to find the financial dominance of these index futures market in the Asia Pacific region and hopefully it may be used as an investment decision reference for domestic and foreign investors. The empirical results show that from the total integration and vector error correction model tests and three places all indicate long-run equilibrium stock index futures and short-term interaction. Therefore, these three places can be viewed as a single regional market. In the Granger causality test on the TAIEX futures and Hang Seng Index futures, in spite of TAIEX futures can’t predict Hang Seng Index futures, it is significantly ahead of the CSI 300 index futures. TAIEX futures on the CSI 300 index futures even more impact than the Hang Seng Index Futures. It can explain that the ECFA has been signed and results show closely-related economy. Since the Hang Seng Index futures are mainly from financial and real estate stocks while the mainland-based financial market is mainly from the real economy, Granger predicts ability of CSI 300 index futures is relatively weak. Another test on the impulse response shows that (1) Hang Seng Index Futures is an independent market and is not affected by shocks from Taiwan and the mainland index futures markets, (2) CSI 300 index futures is affected by shocks from Hong Kong and Taiwan because of the gradually open financial markets, and (3) TAIEX futures can be seen as a potential Taiwanese dish economy because it is most vulnerable to other market influences among the three places. To sum up, the forecast variance decomposition tests show that TAIEX futures and the CSI 300 stock index futures are vulnerable to fluctuations in the Hang Seng index futures. In order words, the Hang Seng Index futures have the strongest explanatory power among the three places and shows financial dominance. The TAIEX futures also show its significant impact on the mainland China financial futures index. If the Government decides to promote the operation of the Asia-Pacific financial center and to increase competitiveness of Taiwan stock market, it will inevitably have to strengthen inter-regional integration efforts and make the most favorable policies to attract foreign investment.

過去已有許多技術應用來建立預測財務危機的模型，如統計學的多變量分析或是類神經網路等分類技術。這些早期預測財務危機的模型大多以財務比率作為變數。然而歷經安隆（Enron）、世界通訊（WorldCom）等世紀騙局，顯示財務數字計算而成的財務比率有其天生的限制，無法在公司管理階層蓄意虛增盈餘時，及時給予警訊。因此，本論文初步探勘共同比分析、公司治理及傳統的Altman財務比率等研究方法，試圖突破財務比率在財務危機預測問題的限制，選出可能提高財務危機預測的特徵群。接著，我們進一步應用基因演算法篩選質性與非質性的特徵，期望藉由基因演算法裡子代獲得親代間最優基因的交配過程，可以讓子代的適應值最大化，找出最佳組合的特徵群，然後以此特徵群訓練支持向量機預測模型，以提高財務預測效果並降低公眾的損失。實驗結果顯示，共同比分析與公司治理等相關特徵確實能提升預測財務危機模型的預測效果，我們應當用基因演算法嘗試更多質性與非質性的特徵組合，及早預警財務危機公司以降低社會成本。

表面增強雷射脫附遊離/飛行時間質譜(SELDI-TOF-MS)是種屬於高維度的蛋白質質譜儀資料，主要是用來偵測蛋白質分子的表現。由於SELDI技術的限制，導致掃描出來的質譜儀資料往往存在誤差與雜訊，因此在分析前通常會先針對原始資料進行低階的事前處理，步驟包括去除基線、正規化、峰偵測(peak detection)與峰調準(peak alignment)。本文中所探討前列腺癌資料，可分成正常、良性腫瘤、癌症初期與癌症末期四種類別。我們分析及比較兩筆事前處理的蛋白質質譜資料，包括我們自行處理的以及Adam等人所處理的資料。為了解決SELDI在偵測分子質量時常出現的位移誤差以及同位素的問題，我們提出以”質荷比段落”當作新的特徵變數的想法來進行分析。本文利用「ROC曲線下面積」(AUC)當作選取的準則來挑選出重要的質荷比段落，而分類方法則採用支援向量機(SVM)。在四分類的分類結果中，我們自行處理的事前處理資可以得到訓練資料89%及測試資料63 %的正確率。而Adam等人所處理的事前處理資料，則得到訓練資料94%及測試資料86 %的正確率。本研究結果指出不同事前處理的方法對分類結果確實有影響，同時也驗證了利用”特徵變數段落”的方法來進行分析的可行性。 / The surface enhanced laser desorption/ionization time-of-flight mass spectrometry (SELDI-TOF MS) is a technique for presenting the expression of molecular masses. It is obvious that every spectrum has a huge dimension of features. In order to analyze these types of spectra samples, preprocessing steps are necessary. The steps of preprocessing include baseline subtraction, normalization, peak detection, and alignment. In our study, we use a prostate cancer data for demonstration. This prostate cancer data can be classified into four categories, namely, healthy men, benign prostate hyperplasia, early stage prostate cancer, and late stage prostate cancer. We analyzed both the preprocessed data processed by ourselves and the preprocessed data done by Adam et al.. In this thesis, we use segmentations of features as “new features” in attempt to solve problems due to location shifts and isotopes. The selection of important segmentations was based on the values of AUC and the SVM was applied for classification. For four-class classification, 94 % and 86 % of accuracy were obtained for training samples and validation samples, respectively, by using Dr. Adam et al.’s preprocessed data, and 89% for training samples, and 63% for validation samples by using our preprocessed data. This study suggested that the preprocessed method does have effect on classification result and a reasonable classification result can be obtained by using segmentations of features.

藉由「早期發現，早期治療」的方式，我們可以降低癌症的死亡率。因此找出與癌症病變有關的生物標記以期及早發現與治療是一項重要的工作。本研究分析了包含正常人以及攝護腺癌症病人實際的蛋白質質譜資料，而這些蛋白質質譜資料是來自於表面強化雷射解吸電離飛行質譜技術（SELDI-TOF MS）的蛋白質晶片實驗。表面增強雷射脫附遊離飛行時間質譜技術可有效地留存生物樣本的蛋白質特徵。如果沒有經過適當的事前處理步驟以消除實驗雜訊，ㄧ 個質譜中可能包含多於數百或數千的特徵變數。為了加速對於可能的蛋白質生物標記的搜尋，我們只考慮可以區分癌症病人與正常人的特徵變數。 基因演算法是一種類似生物基因演化的總體最佳化搜尋機制，它可以有效地在高維度空間中去尋找可能的最佳解。本研究中，我們利用仿基因演算法(GAL)進行蛋白質的特徵選取以區分癌症病人與正常人。另外，我們提出兩種兩階段仿基因演算法(TSGAL)，以嘗試改善仿基因演算法的缺點。 / Early detection and diagnosis can effectively reduce the mortality of cancer. The discovery of biomarkers for the early detection and diagnosis of cancer is thus an important task. In this study, a real proteomic spectra data set of prostate cancer patients and normal patients was analyzed. The data were collected from a Surface-Enhanced Laser Desorption/Ionization Time-Of-Flight Mass Spectrometry (SELDI-TOF MS) experiment. The SELDI-TOF MS technology captures protein features in a biological sample. Without suitable pre-processing steps to remove experimental noise, a mass spectrum could consists of more than hundreds or thousands of peaks. To narrow down the search for possible protein biomarkers, only those features that can distinguish between cancer and normal patients are selected. Genetic Algorithm (GA) is a global optimization procedure that uses an analogy of the genetic evolution of biological organisms. It’s shown that GA is effective in searching complex high-dimensional space. In this study, we consider GA-Like algorithm (GAL) for feature selection on proteomic spectra data in classifying prostate cancer patients from normal patients. In addition, we propose two types of Two-Stage GAL algorithm (TSGAL) to improve the GAL.

癌症至今已連續蟬聯並高居國人十大死因之首，由於癌症初期病患接受適時治療的存活率較高，因此若能「早期發現，早期診斷，早期治療」則可降低死亡率。本文所引用的資料庫，是經由「表面強化雷射解吸電離飛行質譜技術」（SELDI-TOF-MS）所擷取建置的蛋白質質譜儀資料，包括兩筆高維度資料：一筆為攝護腺癌症，另一筆則為頭頸癌症。然而蛋白質質譜儀資料常因維度變數繁雜眾多，對於資料的存取容量及運算時間而言，往往造成相當沉重的負擔與不便；有鑑於此，本文之目的即在探討將高維度資料經由維度縮減後，找出分錯率最小化之分析方法，希冀提高癌症病例資料分類的準確性。 本研究分為實驗組及對照組兩部分，實驗組是以主成份分析（Principal Component Analysis，PCA）進行維度縮減，再利用支持向量機（Support Vector Machine，SVM）予以分類，最後藉由重疊法（Overlap）以期改善分類效果；對照組則是以支持向量機直接進行分類。分析結果顯示，重疊法對於攝護腺癌症具有顯著的改善效果，但對於頭頸癌症的改善效果卻不明顯。此外，本研究也探討關於蛋白質質譜儀資料之質量範圍，藉以確認專家學者所建議的質量範圍是否與分析結果相互一致。在攝護腺癌症中的原始資料，專家學者所建議的質量範圍以外，似乎仍隱藏著重要的相關資訊；在頭頸癌症中的原始資料，專家學者所建議的質量範圍以外，對於研究分析而言則並沒有實質上的幫助。 / Cancer has been the number one leading cause of death in Taiwan for the past 24 years. Early detection of this disease would significantly reduce the mortality rate. The database adopted in this study is from the Protein Mass Spectrometry Data Sets acquired and established by “Surface-Enhanced Laser Desorption/Ionization Time-of-Flight Mass Spectrometry” (SELDI-TOF-MS) technique, including the Prostate Cancer and Head/Neck Cancer Data Sets. However, because of its high dimensionality, dealing the analysis of the raw data is not easy. Therefore, the purpose of this thesis is to search a feasible method, putting the dimension reduction and minimizing classification errors in the same time. The data sets are separated into the experimental and controlled groups. The first step of the experimental group is to use dimension reduction by Principal Component Analysis (PCA), following by Support Vector Machine (SVM) for classification, and finally Overlap Method is used to reduce classification errors. For comparison, the controlled group uses SVM for classification. The empirical results indicate that the improvement of Overlap Method is significant in the Prostate Cancer case, but not in that of the Head/Neck case. We also study data range suggested according to the expert opinions. We find that there is information hidden outside the data range suggested by the experts in the Prostate Cancer case, but not in the Head/Neck case.

領域詞典包含許多專業的詞彙以及對詞彙的定義，但詞典中詞彙間的關係是被隱藏起來的，本研究運用自然語言處理的相關技術，提出運用領域詞典找出詞彙間關係建構特定領域語義網路的方法。 / A domain dictionary contains many professional words and their definitions. In general, there are many hidden relations among words in a dictionary. In this thesis, we use techniques of natural language processing to find out these relations, and bring up a method to construct a domain specific lexical semantic network.

自行車一詞儼然成為綠色環保的代名詞之一，台灣自行車業過去在國際間，被認定為品質粗糙的產品，在經過多年努力的情況下，台灣自行車業不斷的備受肯定，隨著近年全球暖化議題、全球性健康概念、油價飆漲、金融海嘯爆發等，諸多原因造成自行車從不被看好的代步工具，演變到現在成為休閒運動工具的轉變，其中；台灣自行車2008年的金融海嘯中，相較於其他傳統產業，不論是出口產值或是股價不降反漲，大舉逆勢成長，其中巨大(Giant)、美利達(Merida)、愛地雅(Ideal)，成車製造商，近年來分別占出口前三大。 所以本研究要探討，金融海嘯爆發的前後，對台灣自行車業帶來的影響，研究資料選定為2000年1月至2013年12月間的巨大股價(9921)、美利達股價(9914)、愛地雅股價(8933)、台灣股價加權指數(TWII)、原油價格、工業生產指數的月資料，共168筆。透過單根檢定檢測資料是否為定態，利用共整合檢定確定是否含有至少一組解，搭配向量誤差修正模型檢測變數間的長短其關係，在利用複迴歸模型檢測。 研究結果顯示，巨大、美利達、愛地雅和台灣加權股價指數具有顯著關係，由於台灣自行車屬於出口導向以及中高價位產品，故全球景氣對台灣自行車業深具影響。其中，巨大和美利達除了ODM外，亦有自有品牌在全球銷售，愛地雅定位專業ODM專業代工廠，前者發展不同市場。 / The word "bicycle" has become one of the pronouns of environmental protection. In the past, Taiwan bicycling industry was treated as low-quality products internationally. With long-time effort, Taiwan bicycling industry was highly appreciated. Recently, global warming issue, cosmopolitan health sense, dramatically increased oil price, the eruption of financial crisis, and many reasons lead the bicycles have not positively evaluated as means of transportation. Now, it becomes the outdoor recreation mean. Comparing Taiwan bicycling industry with other traditional industry, it doesn't fall down but highly increase no matter export value or stock price. The manufacturer of Giant, Merida, and Ideal are the top 3 of export recently. So this study want to explore the things happened before and after the outbreak of the financial crisis that affects bicycle industry in Taiwan, research data for selected between January 2000 and December 2013, relationship between the Giant(9921) shares, Merida (9914) shares, Ideal(8933) shares, TWII, the price of crude oil, industrial production index. Through the Unit Root Test to test whether the data is the steady state or not. By using cointegration test to make sure whether contains at least one group of solutions and vector error correction model to detect the length of the relationship between variables, and using the multiple regression model to test. Results of the research shows that Giant, Merida, Ideal has significant relationship with TWII, because Taiwan bicycle are export-oriented and high price products, so the global boom has profound influence to Taiwan bicycle industry, among them, the Giant and Merida except the ODM, have their own brands in global sales, Ideal professional locate, ODM professional contract, the former develops different markets. / 摘要 I Abstract II 謝辭 III 目錄 IV 圖目錄 VI 表目錄 VII 第一章　緒論 1 第一節　研究動機 1 第二節　研究目的 3 第三節　巨大機械工業股份有限公司簡介 4 第四節　美利達工業股份有限公司簡介 5 第五節　愛地雅工業股份有限公司簡介 6 第六節　研究架構 7 第二章　文獻回顧 9 第一節　國內相關文獻 9 第二節　國外相關文獻 11 第三節　國內外文獻一覽表 12 第三章　研究方法 20 第一節　單根檢定 20 第二節　共整合檢定 22 第三節　向量誤差修正模型(VECM) 24 第四節　迴歸分析 24 第四章　實證分析 26 第一節　資料來源與處理 26 第二節　敘述統計 31 第三節　單根檢定 32 第四節　共整合檢定 33 第五節　向量誤差修正模型(VECM) 33 第六節　複迴歸模型 35 第五章　結果分析與建議 38 第一節　結果分析 38 第二節　建議 39 參考文獻 40 附錄一　巨大工業股份有限公司沿革 43 附錄二　美利達股份有限公司沿革 47 附錄三　愛地雅股份有限公司沿革 57 圖目錄 圖1-6 研究架構 8 圖4-1-1 台灣自行車業總出口產值(百萬元，美金) 27 圖4-1-2 台灣股價大盤指數(TWII，當日收盤價) 27 圖4-1-3 巨大股價(9921，當日收盤價) 28 圖4-1-4 美利達股價(9914，當日收盤價) 28 圖4-1-5 愛地雅股價(8933，當日收盤價) 29 圖4-1-6 國際原油價格(西德州，美元) 29 圖4-1-7 台灣工業生產指數 30 表目錄 表1-3 巨大公司基本資料 4 表1-4 美利達公司基本資料 5 表1-5 愛地雅公司基本資料 6 表2-3 國內外相關文獻整理 12 表4-1 資料來源一覽表 26 表4-3-1 ADF 單根檢定 32 表4-3-2 單根檢定-一階差分 32 表4-4-1 共整合檢定 33 表4-5-1 Giant & Merida　向量誤差修正模型 34 表4-5-2 Giant & Ideal　向量誤差修正模型 34 表4-5-3 Merida & Ideal　向量誤差修正模型 34 表4-6-1 自行車產業與景氣循環對巨大股價之影響 37 表4-6-2 自行車產業與景氣循環對美利達股價之影響 37 表4-6-3 自行車產業與景氣循環對愛地雅股價之影響 37